Affect and the perception of injustice

Steven J. Scher
Department of Psychology
University College of the Cariboo
P.O. Box 3010
Kamloops, B.C., Canada, V2C 5N3
(604) 371-5517

David R. Heise
Department of Sociology
744 Ballantine Hall
Indiana University
Bloomington, In. 47401
(812) 855-3019

October 23, 1992
Pre-publication draft. For final text see:
Steven J. Scher and D. Heise, "Affect and the perception of injustice." Advances in Group Process, 10 (1993): 223-252.


Traditional approaches to distributive justice have seen the determination of whether or not a distribution of rewards is fair as a cognitive process, with emotion entering the process only as an outcome of a decision that the distribution was unjust. In this paper, we propose a modification of this view. Namely, we propose that justice is not calculated unless the actor feels a justice-related emotion (anger or guilt). These emotions, which arise in the course of social interaction, lead to the instigation of justice deliberations. Using Affect Control Theory, we explain how the justice-related emotions could arise in situations that traditional models of justice would characterise as unjust. Thus, our theory is able to account for the existing literature on justice. We then show how our theory suggests several novel implications about situations that would be seen as unjust. Comparisons of our model to related models of justice are also discussed.

Any social interaction can be seen, from an individual's point of view, as a contribution of inputs, and a receipt of outputs. This is clearest in economic relationships where one party pays a second party for doing some work. More controversially, perhaps, we can view friendships and romances in a similar way. Each member in a social relationship makes contributions and receives certain benefits. In a marriage, for example, the husband may regularly "contribute" a certain amount of money, love and affection, house-cleaning, and grocery shopping to the relationship. In turn, he receives from his wife additional money from her salary, her love and affection, cooking, laundry, and child-care. Each member of the dyad contributes and each receives benefits. A group also can be seen from this perspective. Each group member contributes to the mutual goals of the group (even if the goal is merely continued existence), and in return, each member of the group receives certain benefits of membership.

An important corollary of this exchange perspective is that the motivation of individuals to remain in the relationship (i.e. to keep the job, to stay in the marriage, or to remain a member of the group) will depend on the extent to which they perceive their rewards as fair or just.[1] Perception of unfair rewards can lead to termination of relationships, or dissolution of groups. It also may lead to rebellions in which control of the distribution of rewards is removed from the current allocators, in order to establish procedures for allocating rewards that are perceived as more just. Thus, the study of the causes and consequences of fairness judgments—the study of distributive justice—becomes central for understanding whether social relationships live or die and whether economic and political structures engender satisfaction or unrest and conflict.

Of course, the importance of these types of judgments has not escaped the notice of sociologists and psychologists interested in group processes and interpersonal behavior. Research and theory in this area has progressed over three decades from relatively simple views of how people decide whether or not a distribution is just, to formal and increasingly sophisticated models of the distributive justice process.

The earliest influential theories of distributive justice were the equity theories (Adams, 1965; Homans, 1961; Walster, Berscheid, & Walster, 1973, Walster, Walster, & Berscheid, 1978). Homans' (1961) simple formula for justice stressed the difference between the rewards (R) people receive for some input (I). Adams' (1965) more influential theory of equity proposed that a distribution would be considered just to the extent that the ratio of an individual's rewards to his or her inputs (or investments) was equal to a similar ratio for another person (i.e. RA/IA = RB/IB, where subscripts A and B denote two people participating in the distribution). From this simple beginning, various theorists have developed theories of equity, which have encompassed ever-broader issues, and have grown in mathematical complexity.

Jasso (1978), for example, empirically derived a logarithmic function for evaluating justice. Thus,


In this and other specifications, RB represents the reward-level of any standard of comparison, which may or may not be an actual other person. Theta (q) represents a "signature constant" (Jasso, 1980)—a unique weighting factor for each individual.

Although derived empirically, the logarithmic form for the justice evaluation function meets several important theoretical criteria as well (see, e.g. Jasso, 1990). Markovsky (1985) adopted this form and proposed that "injustice experience" (IE)—the subjective or emotional impact of an unjust situation—can be quantified by:


where JI is justice indifference, the degree to which the person evaluating the fairness of a distribution desires justice for the particular comparison unit.[2] From this perspective, the determination of justice begins with a cognitive evaluation of the congruence of one's outcomes with the expected outcomes (i.e. the reward standard). The emotional reaction (IE), and therefore the likelihood of restorative action being taken, is then a function of this "congruence evaluation" and the degree to which the evaluator is indifferent toward justice in the current situation. (For other formal theories of distributive justice, see e.g. Berger et al., 1983; Blalock & Wilken, 1979; Harris, 1976).

These and other current theories of distributive justice all essentially derive from early equity theory[3] and invoke what Jasso (1986) has called the comparative postulate: "The mechanism by which an individual experiences the (instantaneous) sense of distributive justice involves a comparison of two quantities, the individual's actual amount or level of a good and the amount or level regarded by that individual as `just' for him/herself." (pp. 251-252). Variations in how perceptions of justice are modeled have revolved around two major issues: who forms the basis for the comparison, and how that comparison is structured.

Identifying the Reference Standard. The initial equity theories (Adams, 1965; Walster et al., 1973, Walster et al., 1978) postulated that we determine what is fair by comparing some transformation of our outcomes and inputs to the outcomes and inputs of another person in the same situation. Other theorists in distributive justice, however, have noted that there are other sources for the reference standard against which to compare one's situation. For example, Berger et al.'s (1972) "status value theory" allows that a generalized other can serve as a basis for reward expectancy (see also Berger et al., 1983). From this view, the comparison other need not be a real person, but can be some generalization of people who hold certain status characteristics. Comparison with one's own past experiences also has been suggested as a source of the reference standard. As Austin, McGinn, & Susmilch (1980) put it: "Persons, on the basis of their past experiences, form expectancies of what constitutes a satisfying and fair reward for various types of situations and then use these standards for evaluating current relationships" (p. 428). Thibaut and Kelley's (1959) interdependence theory (i.e. "comparison level"), and Stouffer et al.'s (1949) concept of relative deprivation are well-known examples of theories based on comparison with past outcomes.

Finding the Structure. The other debate within the literature focuses on what qualities of the reference standard and of the individual recipient should be compared, and how the information is integrated for the comparison. Formulae such as Markovsky's or Jasso's above describe this process. Notably, these theories describe the evaluation of the fairness of a distribution as a purely cognitive process. For example, Jasso (1980) describes the ratio of actual outcomes to expected outcomes (the "comparison ration") as "an exclusively cognitive magnitude, completely devoid of emotional content" (p. 6). Rewards and reference standards are compared, and a decision is made regarding the fairness of the rewards. "Short of claiming that actors carry out the mathematical operations, it is assumed that they will behave as if they do, whether consciously or not" (Markovsky, 1985, p. 827). Emotion enters the picture only as a result of this cognitive evaluation of justice.

In this paper, we offer a revision of the above theoretical formulations. Our view incorporates a more interactive process, with affective concomitants, and sees justice decisions as affective in origin even if mostly cognitive in form. Theorists of distributive justice have been criticized for ignoring the "interactional character of the exchange relationship" (Deutsch, 1979, p. 30; see also Deutsch, 1983). In a review of several books on social psychological approaches to justice, Morton Deutsch (1983) characterizes the state of the justice literature: "The approach to `justice' has been too psychological and not enough social psychological [sic]; that is, it has focused on the individual rather than upon the social interaction in which 'justice' emerges ... Much of the current work ignores the relationship between conflict and justice and the process by which 'justice' is negotiated." (p. 312).

Our central thesis is that the evaluation of a rewarding act depends crucially on affective responses arising within the interaction, and that these affective responses emerge from the way the actors perceive the role-identities held by the various interactants, the definitions of the various actions, and the ways that these meanings combine in on-going social interaction.

We begin by reviewing research showing that affect does impinge upon attributions of justice, suggesting that these judgments are affect-laden and tied to specific affective conditions and emotions. Next, we will show how a general theory of social interaction would predict that these emotions could be generated in situations that typically produce judgments of injustice. Finally, we will discuss various implications of this approach to distributive justice.

Affect and Justice Judgments

Traditional theories of distributive justice have discussed the emotional effects of unfair distributions as a by-product of the cognitive judgments about justice. For example, Homans (1961) writes "The more to a man's disadvantage the rule of distributive justice fails of realization, the more likely he is to display the emotional behavior we call anger. Distributive justice may, of course, fail in the other direction, to the man's advantage rather than to his disadvantage, and then he may feel guilty rather than angry..." (pp. 75-76). Other researchers have followed Homans' lead to argue that unjust distributions lead to the negative emotions of anger and guilt. The cognitive evaluation of congruency is proposed as a major element in the calculation of the emotional effects of injustice (e.g. Jasso, 1980; Markovsky, 1985). In fact, some have argued that injustice is part of the prototypical script which defines the emotion anger. For example, Lakoff (1987) describes the first event in the prototypical scenario for anger in the following manner: "There is an offending event that displeases S ... the offending event constitutes an injustice and produces anger in S" (p. 397, emphasis added).

On the other hand, recent research has demonstrated that people's judgments about justice can be changed by their affective state. Mark and Sinclair (1992, cited in Sinclair and Mark, 1992) induced either an elated, neutral, or depressed mood in subjects who subsequently read about work situations. The payment in these stories ranged from equality, through equity, to inequitable overpayment. As predicted, pay structures that deviated further and further from equity (toward either equality or overpayment) were rated as increasingly less fair. More importantly, however, the greatest amount of variation occurred among subjects who were in a negative mood, and the least variation occurred among those in a positive mood. In other words, subjects in negative moods saw equality and overpayment as more unfair, relative to equity, than did subjects in positive moods. This supports the notion that subjects in a negative mood are more stringent about what they regard as fair.[4] In a second study, Mark and Sinclair (1992) found that subjects who were in good moods considered a broader range of payment for work done as fair, relative to subjects in bad moods, providing further evidence that justice judgments are linked to affect.

O'Malley and Davies (1984) also manipulated mood and measured subject's allocations of rewards to either themselves or to others. They had subjects engage in a puzzle where performance was quantifiable, and, after a fixed time period, subjects were led to believe that they had done either better or worse than another subject. They then were asked how they thought they should divide 100 raffle tickets between themselves and a partner. For subjects who performed better than the supposed other subject, mood had no effect—all subjects behaved in a fairly selfish way, taking a high percentage of the raffle tickets for themselves. However, for subjects who performed worse than the other, mood did affect the allocation strategies. In particular, subjects in a happy mood took more of the raffle tickets, and subjects in a sad mood took fewer tickets, when compared to subjects in a neutral mood.

The above studies dealt with diffuse moods rather than specific emotions; however, two studies have examined the effect of anger on attributions of blame and intentionality. Gallagher and Clore (1985) used hypnosis to induce either anger or fear in subjects, and then had them read a story about a transgression. Angry subjects were more likely to attribute blame to the transgressor than fearful subjects. Melton and Scher (1992) asked subjects to recall in detail an incident which had made them angry. Even though these subjects were most likely aware of the source of their anger, they were more likely to attribute intentionality to a transgressor they read about, than were subjects who had recalled a happy incident or those who had recalled a non-emotional incident. These studies suggest that people's moods can influence their interpretations of justice and that judgments relevant to justice can be influenced by specific emotions.

Indeed, there is a growing body of research on the effects of mood or emotional state on cognitive processes (see Isen, 1987; Schwarz, 1990; Sinclair & Mark, 1992 for reviews). People sometimes use their current emotional state as a source of information when making judgments (see Schwarz, 1990; Schwarz and Clore, 1988). People also use others' emotional states as information in making judgments about these others (Heise, 1989). Making inferences from how a person is feeling can serve as a useful heuristic method of arriving at decisions about complex issues.

Norbert Schwarz and Gerald Clore have demonstrated that when people are asked about their subjective well-being, those who have had their mood enhanced (e.g. by finding a dime on a copy machine [Schwarz, 1983] or by being interviewed on sunny days [Schwarz and Clore, 1983]) report being better off than those whose moods were not elevated (i.e. they did not find a dime, or they were interviewed on rainy days). The given explanation for these and related data is that the subjects were accessing their current affective state as a way to arrive at a decision about their well-being.

However, this effect only occurs when the affective information appears relevant to the judgment to be made. When the affective state was attributed to something else, it no longer supported an inference backwards to well-being, and the effect disappeared. For example, in the weather study (Schwarz and Clore, 1983), some subjects were asked at the beginning of the interview how the weather was in their town (the interviewer was presumably calling from another city). Those subjects who were phoned on a rainy day, and who were asked about the weather, subsequently reported being just as happy and satisfied with their lives as those who were phoned on sunny days. In other words, their mood, on being attributed to the weather, had no effect on their judgments about subjective well being.[5] Consistent data were reported by Schwarz, Servay, and Kumpf (1985). They showed male cigarette smokers a fear-arousing film designed to persuade people to reduce cigarette use. Some of these subjects were given a pill (actually a placebo) and told that the pill had a side effect of arousing those who took it. Other subjects were led to believe that the pill would calm them down, and still others were told that the pill had no side effects[6]. The reasoning was that those who believed the pill was arousing them would be less likely to treat their fear as information in deciding whether to try and cut down on cigarette use, because they would attribute any arousal they felt from watching the film to the pill. On the other hand, those subjects who believed that they should be calmer because of the pill should be even more inclined to reduce their smoking. This is indeed what happened (although non-significantly so for the comparison between the "arousing pill" condition and the "no-side effects" condition).

These studies show that justice-related judgments and certain other reasoned decisions are influenced by affective states. In this paper, we go a step further and propose that the emotions associated with unjust distributions of reward—anger and guilt—may be aroused interactionally in situations where justice concerns are at issue. These emotions may then guide interactants to judgments about events being unfair, rather than vice versa. Characteristics of the interaction may lead interactants to feel angry (or guilty), and, in searching for a way to account for their predicament, they may draw conclusions regarding justice from the information offered by their emotional state. The information offered by the presence of anger is contained in the prototypical script for anger –-that another acting unjustly toward us produces anger. Ergo, feeling anger in an interaction implies one's interaction partner acted unjustly. The information offered by the presence of guilt is contained in the prototypical script for guilt (Kemper, 1990, p. 223) –-that our unjust denial of what is due to others produces guilt. Ergo, feeling guilt in an interaction implies we dealt with someone unjustly.

However, although these culturally-defined scripts for emotions can be accessed for information regarding justice-based decisions, this does not necessarily mean that these scripts do indeed describe the causes of the emotions. We must distinguish between the processes by which an emotion is aroused, and the prototypical definition of the concept held by members of a culture (Russell, 1991). Anger and guilt may arise out of normal social processes within an exchange transaction, but still provide information as if the emotion was caused by justice concerns.

Our proposition is this: When people involved in a transaction feel anger or guilt, and the emotions are not ameliorated, they may decide that the transaction is unfair or unjust. The scope is limited to transactions (exchanges) because that is where the concepts of justice or fairness are relevant, and people ordinarily would not employ these concepts to understand their emotions in other kinds of relationships. Alternative resolutions of the emotions within the transaction may also take place; anger and guilt sometimes can be reduced within transactions so as to preclude the need for managing them.

Our position relates unresolved justice-related emotions to Markovsky's (1985) concept of justice indifference. According to Markovsky's formula for injustice experience (see Equation 2, above), the degree of justice indifference determines the emotional impact of a given incongruence. As indifference approaches infinity (i.e. complete indifference), injustice experience approaches zero—there is no emotional impact, no matter how large the incongruence. On the other hand, as indifference approaches its lower bound (in this case, JI=1), injustice experience increases. Put another way, the more one cares about injustice, the greater the emotional impact of a given incongruence.

Our proposition also proposes a factor which influences the degree to which a particular incongruence will result in a subjective (affective) experience and justice restoring behavior. This factor is justice-related emotion. However, the route by which the consequences of injustice are modulated by affect is different from the route proposed by Markovsky for justice indifference. Specifically, Markovsky's formulation suggests that the cognitive evaluation of incongruence is made no matter what, but that the effect of that evaluation is reduced when justice indifference is high. In contrast, we propose that, if justice-related affect is absent, no evaluation of incongruence (i.e. of injustice) will be made. That is, increases in the intensity of the affect do not increase the effect of a given incongruence, but rather the likelihood that the degree of congruence will be `calculated' at all.

For our proposition to be interesting, we must show that stereotypically unfair interactions can indeed produce justice-related emotions before participants make an actual judgment of unfairness. This would demonstrate that our proposition might be an adequate way of describing how justice judgments arise, and that "affect leads to justice-judgments" is just as plausible as "justice-judgments lead to affect". Yet our proposition still is no more than an exotic alternative to standard views unless we meet a second condition, showing that our sociology of emotion approach adds something new to the computational approach to justice. Toward this end, we will suggest how some seemingly unfair transactions might be perpetuated without judgments of unfairness, and how some seemingly benign transactions might instigate justice deliberations.

In order to meet these goals, we must first adopt a model of social interaction. We will use affect control theory (Heise, 1979; MacKinnon & Heise, forthcoming; Smith-Lovin & Heise, 1987) to cast the distributive justice process into social interactional terms, because the current work arose from trying to examine distributive justice in terms of that theory, and because we feel that affect control theory is the most thoroughly developed theory to account for the ever-changing meanings, actions, and feelings that make up social interaction. After an introduction to affect control theory, we will use the theory to show how social interaction could produce justice-related emotions that induce judgments of justice.

Affect Control Theory [7]

Affect control theory is so named because it posits that "the basic motivational principle ... is that people construct or reconstruct events so as to maintain consistency between transient feelings and sentiments." (MacKinnon and Heise, forthcoming). The meanings applied to actions and actors in a social interaction, and the behaviors taken by an actor, are directed toward `controlling' the discrepancy between an observer's pre-existing sentiments about the elements of the interaction and the transient impressions created by specific acts.

The theory uses Osgood's (1962) three dimensions of affective meaning to measure sentiments and impressions. Evaluation is the familiar attitudinal component of response—the feeling that something is good or nice as opposed to bad or awful. The Potency dimension of response registers the sense that something is powerful versus powerless. The Activity dimension relates to assessments of whether something seems lively and fast, or whether it seems quiet and slow. Research conducted in more than twenty nations (Osgood, May, and Miron, 1975) indicates that evaluation, potency, and activity (EPA) are universal dimensions of response to stimuli of many kinds, and that sentiments vary along these dimensions cross-culturally. Individuals, behaviors, and settings all can be rated on these dimensions, and the averaged ratings index culturally-defined sentiments about these social objects. Within a culture, mean EPA ratings for most stimuli are very similar for males and females and for people of different socioeconomic levels from various regions of the U.S. (Heise, 1966). Subcultures support divergent sentiments but only for concepts that are very central to the subculture's routines (Heise, 1979: 100-102; Smith-Lovin and Douglass, forthcoming).

Predicting Transient Impressions. Interpersonal events create transient impressions of people, behaviors and settings, and these impressions can differ from the pre-existing sentiments. One paradigm for studying the impressions created by events (Gollob, 1968; Gollob and Rossman, 1973; Heise, 1969, 1970) is to present an event (e.g., The employee neglects the employer), then ask subjects to rate the actor (e.g., the employee) in the context of the event. Averaging responses across subjects yields an EPA profile assessing the impression produced by the actor engaging in the given behavior with the particular object person—an impression that usually is different from the sentiment (measured out of context) that represents what we think of the actor generally. Alternatively, subjects may be asked to rate the object person (the employer) or the behavior (neglecting), given the context of the event.

Transient impressions can be predicted quite accurately from EPA mean ratings of the actor, behavior, and object-person obtained outside an event (Gollob, 1968; Gollob and Rossman, 1973; Heise, 1969, 1970, 1979; Heise and Smith-Lovin, 1981; MacKinnon, forthcoming; Smith-Lovin and Heise, 1982; Smith-Lovin, 1987a, 1987b). The predictive equations come from regression analyses in which in-context EPA ratings of event elements are regressed on out-of-context EPA ratings of the event elements (e.g., an employee, an employer, to neglect someone) plus multiplicative interaction terms. Gollob (1968) and Heise (1979, chapter 2), note that some of the interaction terms correspond to classical notions of attitude balance.

Here, for example, is the equation based on a regression analysis across 515 events (Smith-Lovin, 1987a) for predicting the transient impression of a behavior on the evaluative dimension:


A, B, and O refer to the actor, behavior and object-person, respectively; the subscripts e, p, and a refer to the evaluation, potency and activity dimensions. The transient impression (the dependent variable) is denoted with a primed symbol; feelings existing before the event (the predictor variables) are denoted without primed symbols. Only significant predictors are shown.[8]

The terms in this equation can be interpreted as follows (Smith-Lovin, 1987a: 47-49). First, evaluation of the actor (Ae) is a factor determining impression of the behavior. Any behavior seems slightly better when done by a good actor or slightly worse when done by a bad actor. The coefficient of the second term (Be) reflects a substantial degree of stability in impression formation: good acts preserve some of their goodness or their badness regardless of the circumstances of their use. The presence of the other behavior terms (Bp, Ba) can be viewed as meaning that a behavior's evaluative stability is especially great for weak and quiet acts. The identity of the interaction partner (Oe) also reflects on the behavior to a small degree, in the same manner as the identity of the actor does.

The AeBe term represents an actor-behavior consistency effect (Gollob, 1968). People evaluate behaviors more negatively when they are evaluatively out of character for the actor. For example, sexual abuse is seen as worse if it is carried out by a mother than if it is carried out by a deviant. Behavior-object consistency is another force affecting the evaluation of a behavior. The BeOe effect shows that behaviors are evaluated more positively when they are directed at evaluatively appropriate object persons (e.g., helping the deserving, or rebuking deviants). The BpOp effect means that powerful behaviors are evaluat­ed more positively if they are directed toward strong people, and more negatively if directed toward weak objects.

The interaction of actor evaluation and behavior potency (AeBp) shows that powerful behaviors (like challenge) seem better when conducted by nice actors while weak, power­less acts (e.g., imitate, beg) seem better when done by disvalued actors.

The BeOp interaction originally was identified by Gollob and Rossman (1973) in a study of actor potency. Directing good acts at a weak person (e.g., aiding a victim) or bad acts at a powerful person (criticizing a president) makes people view the action more positively. By contrast, we disdain acts of ingratiation with the powerful or of nastiness toward the weak. The behavior potency-object evaluation interaction (BpOe) is conceptually similar to the BeOp effect. Here again, the evaluation dynamics seem to reflect a "just world" assump­tion. Behaviors seem better if they represent gentle treatments of good objects or forceful treatments of bad objects. Conversely, a behavior seems worse if it amounts to treating a bad person in a gentle manner or a good person in a dominating manner.

The three–way interactions constitute qualifications of the two–way interactions. That is, the AeBeOe, ApBpOp, and AeBpOp interactions show that the consistency effects, BeOe and BpOp, operate most strongly when the actor is positively evaluated and powerful.

Smith-Lovin (1987a; 1987b) presented similar equations for predicting potency and activity impressions of behaviors, as well as equations for predicting EPA outcomes for actors, for object persons, and for the settings that are involved in events. As it turns out, the equation for predicting evaluation of an actor is very similar to the above equation for predicting evaluation of the actor's behavior, except the coefficient for Ae is much larger and the coefficient for Be is much smaller. Impression-formation equations account for about 80 percent (or more) of the variance in transient impressions. Moreover, the equations are similar for subjects in different socioeconomic and national populations to a large degree. (See Smith-Lovin, 1987a, for a review of results from impression-change studies with U.S., Irish, and Arabic subjects; MacKinnon, forthcoming, provides results from Canadian subjects.) We therefore assume that the equations represent quite general, ubiquitous aspects of affective reactions to social events.

An important variant of impression-formation research develops equations for predicting outcome impressions when identities are combined with emotions, traits, or status-characteristics (Averett and Heise, 1987; Heise and Thomas, 1989). These equations allow affect control theory to deal with situated identities—e.g., an angry father, a wise child, a rich professor—and the equations also are the basis for theoretical modeling of emotional responses and trait attributions.

Emotions. According to affect control theory, emotions arise as an indication of how well the current situation is confirming the expected identities of interactants. Emotions are temporary affective conditions that register how events are making one seem as compared to how one is supposed to be (cf. Higgins, 1987 for a similar theory of emotion). A person invokes an emotion that combines with his or her identity within a situation to generate a transient impression identical to the transient impression created by the current event. That is, the impression of an angry father is less good than for just father. So, if a father is involved in an event which makes him appear less good, he will become angry (i.e. he will make himself angry) in an effort to feel the impression he has created. Thereby, the person viscerally experiences how impressions created by the event relate to his or her identity. By knowing both the situational identity of an individual, and the transient impression of self created by an event, one can predict the emotions that are likely. The prediction equations are obtained from the empirical equations that predict impressions of emotion-identity combinations (like "angry father"), given the profiles for the identity (father) and the emotion (angry). Solving the equations for emotion yields new equations that predict emotion from identity and impression[9].

The model predicts that emotions directly correspond to how events have affected the self—an interpretation that corresponds to intuitions (e.g., events that make one look bad also make one feel bad). However, situational identities influence emotions in several ways. In the first place, people conduct themselves so as to keep transient impressions of themselves close to their identities, according to the basic axiom of affect control theory. Therefore identity determines emotion by determining what transient impressions generally arise as an individual creates events. Additionally, the model indicates that emotions reflect how transient impressions of self compare to one's identity. This suggests, for example, that people experience especially good or potent or lively emotions when events make them seem more good or potent or lively than their identity warrants. Finally, an interaction effect in the equations indicates that the evaluation of one's situational identity influences the extent to which transient impressions of self translate into more extreme emotions; one consequence of this is that people with extremely negative self identities may experience chaotic emotions, or emotional lability.

The Course of Interaction. For the purposes of affect-control analyses, the basic progression in a social encounter is as follows:

First, participants comprehend social situations by recognizing people in terms of personalized identities (e.g., John, Mary), informal social stereotypes (e.g., he-man, maverick), or formal social roles (e.g., secretary, executive). The sentiments for these identities are evoked, and the sentiments serve as affective reference points for understanding each participant in the situation.

Second, participants anticipate, perform, and interpret events to confirm pre-existing sentiments as much as possible. Thus, people with valued identities are expected to perform acts which create positive impressions, whereas interactants with disvalued identities are expected to behave in ways that confirm the stigma that is associated with them.

Third, expressive displays reveal how each interactant is faring in social interaction—whether the interactant seems good, potent, and lively and whether the levels of goodness, potency, and liveliness are appropriate to the interactant's identity. Positive emotion occurs when an interactant benefits from events—especially when the benefit is greater than the person's identity warrants, and negative emotion occurs when events create negative impressions of an interactant—especially when impressions are less favorable than is warranted by the interactant's identity.

Fourth, occurrences that disconfirm identities, and that are not repaired by subsequent restorative events, may cause reassessments in which identities are adjusted to fit the events that have occurred. For example, a person who is sullied by an event may move into a less valued identity or may adjust the current identity by amalgamating it with a negative mood, trait, or status characteristic.

The Affective Nature of Justice Deliberations

In this paper we propose that an additional phenomenon can occur when people try to cope with certain emotions in the context of transactions, especially when the opportunities for restorative actions are very limited, as in the economic or political spheres. In many (if not most) interactions, people have a wide repertoire of behavioral and cognitive means to try and restore discrepancies between event-created sentiments and more fundamental sentiments. But, in some circumstances, including many exchange transactions, these options become limited. When unresolved anger or guilt is produced in these types of interactions, one possible way to deal with the emotion is to characterize the transaction as unjust, opening up the opportunity for

Figure 1. Schematic diagram of the role of emotions in transactions.

restorative action outside of the transaction (e.g. retribution, reparation, or even dissolution of the relationship).

However, the emotion alone is not a sufficient cause for deciding that a transaction is unjust. We propose that justice-related emotions initiate an assessment of an individual's rewards relative to expected rewards, as modeled by traditional justice researchers. If this rationalistic assessment confirms a conclusion that the outcome was unfair, then the victim has the rhetorical ammunition to demand that the problem be resolved via extra-transactional means. However, the assessment may conclude that recent events are just and do not warrant extra-transactional restorative events that would eliminate the emotions. In that case, the emoter is left with stressful emotions that cannot be resolved in the relationship that generated them, and the only recourse is to turn to stress management techniques (discussed, for example, by Thoits, 1990). Our model of how emotion enters justice judgments is shown in the accompanying schematic diagram (Figure 1).

We believe that more interpersonal options available in an interaction allow emotions to be resolved interactionally, and on the other hand that constraint in transactions makes it more likely that justice deliberations will be needed. For example, at the extreme, all a worker can do is work, and any resolution of emotions generated by work depend on actions by the employer. And, at the extreme, all an employer can do is pay, so alleviation of worker emotions depends on how often (or how much) the employer pays and on the meaning of the payment for the worker. The more constrained an interaction is in this way, the less able the parties are to manage emotions interactionally, and the more likely that concerns about fairness will arise.

The crux of our argument is that justice-related emotions can arise in transactions apart from considerations of justice. We will elaborate on this matter in the next section. Here we point out that such emotions have impact because they typically are indicators of disconfirmed identities, and it is the disconfirmation of identity that actually motivates the search for a resolution. Sometimes, however, justice-related emotions will be consistent with one's identity. In situations where anger or guilt do not reflect disconfirmations of identity, we do not expect them to trigger justice deliberations. For example, viewing a mugging as a kind of transaction, the mugger could experience anger in confirmation of his mugger identity, and that anger would not lead him to deliberate about the justice of the transaction. The same emotion, however, experienced by the victim, could, indeed, initiate such deliberations.

We allow that justice-related emotions that arise in a transaction also can be resolved in the transaction, as illustrated in the next section. Such a resolution is a normal outcome, and as long as the transaction diffuses the emotions that it generates, as long as the transaction allows participants to confirm their identities periodically, there is no impetus to turn to justice deliberations.

Examining the justice of a transaction involves searching for meanings of acts—for example, trying to determine the value of payments as judged by comparison with other transactions, or in terms of tradition or other criteria. As mentioned previously, affective states influence these deliberations, but we are not yet ready to model that influence. In this chapter, we simply recognize that a judgment of unfairness both legitimatizes renegotiation of the terms of the transaction and also justifies side payments in the form of retributions or reparations in order to restore parties to their rightful identities.

Ideally a judgment of fairness would adjust the meanings of acts in a transaction so that in retrospect and prospect the identities of parties to the transaction are properly confirmed. That happens sometimes, but less often than seems rational because those experiencing the emotions make use of extra information that disinterested outsiders do not have. That is, a person who is angry or guilty in a transaction "knows" the transaction is unjust because they know what they are feeling, and the prototypical scripts for these emotions have injustice as the initiating condition. The emoter often will trust his or her emotions, and therefore will not be convinced by reasonable arguments that an encounter is fair. As a result, they will not use the fairness judgment to adjust the meanings of acts. Meanwhile, though, they have to recognize that others' fairness judgments do make extra-transactional solutions inappropriate. They end up with the transaction and their troubling emotions intact, and the only resolutions left to them are personal ones like changing their self-concepts in a negative direction, or engaging in fantasy, substance abuse, or denial of what one is feeling (see Thoits, 1990). By this account, some mentally disturbed people can be seen as victims of justice.

Prototypical Transactions

We turn now to examining how justice-related emotions might arise in social interaction, and to showing how such emotions ordinarily are resolved naturally, precluding the need for justice deliberations. Our analysis derives from computer simulations of social interaction based on affect control theory's mathematical model combined with EPA ratings from a U.S. population of males (Heise and Lewis, 1988).

Traditional Factors Affecting Justice. Suppose that two people have taken the roles of employee and employer. The employee is expected to maintain a somewhat good, weak, and quiet identity in the transaction (the mean ratings of "employee" are 0.83, –0.53, –0.18 on evaluation, potency, and activity, respectively - see note 8 above for details about these scales). The employer should maintain a good, very powerful, and quiet identity (0.92, 2.00, –0.68). Now the employee engages in work for the employer, and the action of working for the employer can take on a variety of affective meanings, depending on the nature of the work.

In order for our approach to be viable, we need to show that as conditions become more "unjust" (that is, as conditions become such that models of justice would predict judgments of unfairness), the justice-related emotions (anger and guilt) become more likely. If the work being done is not particularly onerous or physically demanding, neither traditional views nor intuitive beliefs would expect judgments of unfairness. Our simulations indicate that under these kinds of work conditions, the employee will not experience anger as a result of performing duties. Therefore, justice would not be considered. By changing the affective meaning of work (i.e. by experimenting with different EPA profiles), we found that anger will arise if the work is even a little unpleasant (say, –0.50 on the Evaluation dimension) and if it is at least slightly active (say, 1.00 on the Activity dimension). The potency of the work, which might be interpreted as its importance, is not crucial—people can get fed up with important work as well as unimportant work—so assume heuristically that the employee views the work as somewhat important (1.00 on the Potency dimension).

Affect control theory predicts that an employee repetitively engaging in such work comes to see the self as slightly bad, potent, and active (–0.76, 0.35, 0.69). The bad impression arises because the employee is a person in a good role trying to relate to another person in a good role through unpleasant action (it would make more sense to relate to the employer through positive action, such as asking the employer about something, or advising the employer—as predicted from affect control theory). Appearing somewhat bad, potent, and active already is one step toward anger (which has an EPA profile of –0.86, 0.21, 0.71). But comparing the impression of self that the employee is creating with the profile for the employee identity reveals that, relatively speaking, the employee seems substantially less good, more potent, and more lively than should be the case. Such a disconfirmation of identity rouses the emotion of anger.

If the employer understands the work in the same way as the employee, then the employer is just as stressed by the situation as the employee. And the employer is able to do something about it within the rules of a transaction. Paying the employee helps bring the employee's self impression back to where it is supposed to be and greatly dissipates the propensity to anger. Making the payment more positive by paying well can wipe out the negative impression of the employee and the negative emotion. On the other hand, failing to pay or paying poorly while continuing to demand arduous work leaves anger and primes the employee to begin deliberating over the fairness of the transaction. Within traditional models of justice, these conditions would likewise be expected to increase unfairness judgments. But we argue that it is the emotional effect which initiates this process.

Other details of the situation can influence whether anger is generated. The sequence of events is one important factor. For example, work that is only a little unpleasant has to be repeated without reward for some time before it begins to generate anger, according to simulations. The physical environment also is significant according to simulations—for example, a dispiriting setting (like working in a slum) can cause pleasant work to engender anger and thereby considerations of fairness.

ACT makes a number of non-intuitive predictions that would not be likely to be derived from traditional approaches to distributive justice. One of the most important, we feel, is the role played by the identities assigned to actors. This is the topic of the next section.

Identity and Justice Emotions. An important subset of factors affecting the arousal of justice-related emotions involves the identities of the actors. As described above, emotions are signals of the current ('transient') identity of an actor in a situation, as well as the relationship between the current identity and the pre-existing ('fundamental') identity. Thus, one way to view the sequence of events we propose from situational factors (e.g. the amount of pay given, the laboriousness of the work) to emotional arousal, to cognitive deliberations about justice is to see the situational factors generating emotions related to the fundamental and transient identities and the relationship between them.

In the course of interaction, the identity a particular actor takes on can become threatened because the events in the interaction are inconsistent with that identity. One aspect of emotion is to signal such a discrepancy. Because the guiding force in social interaction, according to ACT is to seek to reconfirm the identities of actors, the presence of emotion that is inconsistant with one's identity is an indication that some reconfirming action of cognition should take place. ACT allows for many ways for this to occur (e.g. reconceptualization of the event, additional behavior, redefining of various identities). The new resolution suggested in this paper is the transformation of a justice-related emotion into a justice deliberation and then, if warranted, into a reparation outside the normal course of interaction.

This perspective on how justice related emotions, and therefore justice deliberations, are generated leads to another implication of our view on whether or not a particular situation will be considered unjust (or, more exactly, whether or not justice will be considered at all). The way the participants assign identities for themselves and others influences the degree to which justice -related emotions are aroused. Even pleasant work will induce anger and justice deliberations in an employee who already has decided that he or she is a "dissatisfied employee", and work is more likely to engender anger if it is done for a disrespected person like a "slavedriver". On the other hand, even odious work will generate positive emotions in someone who accepts a debased identity like "slave".

Justice and Guilt. Our simulations suggest that guilt is not a very prevalent basis for justice deliberations. Affect-control simulations suggest that employees who receive lavish payments accept their fate because they are being made to look better than they should be, which generates happiness, and happiness (even when it signals disconfirmation of one's identity) is not an emotion instigating fairness deliberations. If an employee engages in a transactionally deviant act, like cheating the employer, the employee more likely will feel anger than guilt. (The anger could lead to considerations of fairness, but the deliberations might end up treating the theft as justified reparation.) An employer engaging in a deviant action, like over-working an employee, also is likely to have a bout of bad temper rather than guilt accompanying the episode.

The fact that the simulations do not suggest the arousal of guilt from overpayment should not be seen as inconsistent with our view of the justice process. In fact, there has been a persistent finding in the literature on distributive justice that people are less likely to call distributions unjust when they receive more than they expect or feel entitled to (e.g. Messick & Sentis, 1979). This so-called egocentric bias (Messick & Sentis, 1983) would be facilitated by the lack of guilt in situations where it "should" occur.

We did find that an employer accepting work from a "slave" could end up feeling ashamed, and convincing an employer that he or she is treating workers as debased beings might make the employer open to justice deliberations. On the other hand, an employer always has the option within the framework of the transaction of engaging in some small payment that alleviates the employer's guilt, even if dealing with slaves.


Throughout this chapter, our argument has been that judgments about justice, and the behaviors that follow from those judgments, are instigated through the affective dynamics of social interaction. The implications of such a model are many. Anything that affects those dynamics should have a predictable effect upon instigation of justice deliberations. In this section, we will discuss several of the more note-worthy implications of this approach, focusing particularly on those implications which we feel have not been derived from traditional cognitive approaches to distributive justice.

A Situation Defined as Real. Affect control theory was developed in an explicit attempt to model the symbolic interactionist view of social interaction. By casting distributive justice as a social interaction, and modeling justice-oriented reactions in terms of this process, we can take into account the subjective nature of definitions of situations as they relate to justice phenomena.

The way that a particular situation is defined, the role-identities assigned to actors involved in a transaction, and the modification of those identities within the context of the transaction all influence whether or not a justice-related emotion arises. If either party takes on, or is cast into, certain negative identities, the transaction can produce negative emotions that can culminate in justice deliberations. For example, an employer who presents a stigmatized identity (Goffman, 1963; Jones et al, 1984) like homosexual (EPA of –1.52, –0.79, 0.64) is likely to rouse anger in employees as they work for him or her, increasing the likelihood that injustice will be experienced. To our knowledge, no other theoretical perspective has suggested that status of the payer would affect the likelihood that a distribution would be seen as unjust. On the other hand, a party to a transaction who accepts a role that is both stigmatized and powerless is unlikely to develop anger and is unlikely to be discontent except by way of guilt; this is similar to the suggestion by Berger and his colleagues (Berger et al., 1972; Berger et al., 1983) that the reward that a payee expects is derived from his or her status.

The initial identities assigned to interactants are not the only way that people can gain stigmatized identities. One important aspect of affect control theory is that behaviors, and the people who engage in them, are evaluated in the context of their current affective meaning in the eyes of the evaluator. And that meaning is constantly changing. Each act carried out by an actor can change sentiments about that actor. So, an actor can become stigmatized by engaging in a stigmatized act. Imagine a simple social interaction, involving an employer and an employee, with the employee engaged in slightly pleasant work. The interaction could have three steps:

1) The employee works for the employer.

2) The employer ignores the employee, whereupon the employee reidentifies the employer as a "stuffed shirt".

3) The employee continues the work, but now feels angry and begins considering the injustice of the transaction.

This vignette is from a simulation, but it plausibly recounts how transactions sometimes go sour as a result of acts that change the way people define their situation. Previous acts within an interaction can lead to such redefinition, and therefore can affect the initiation of justice deliberations.

Do People Typically View Unjust Distributions as Unjust? One implication of our analysis is that there are many ways to eliminate claims of injustice. One can set up a slave society, and if everyone believes in it, it will work—a true slave understands that even the worst work is gratifying and even a small reward is cause for joy. But exploitation can take a more positive cast, too. Roughly speaking, anything that keeps people happy also keeps them from worrying about injustice. For example, an executive might be advised to put company people in nice settings, give everyone status, allocate frequent rewards even if they are small, and help everyone cover behavioral bloopers that could make others think unkindly of them. An outsider might consider the company's reward structure unjust, but most likely the possibility of injustice will never enter the employees' minds, unless something in the situation makes justice considerations salient to them (such as being asked by a social scientist whether they felt their outcomes were fair). From this perspective, we might think of affect as one factor affecting justice indifference (Markovsky, 1985). As discussed above, this term reflects the degree to which people care at all about justice concerns.

We do believe that the rational calculation of rewards and costs, and the comparison of one's own ratio to others', can set off deliberations about fairness. However, we believe that such calculations initiate fairness considerations by defining the meanings of costs and rewards. That is, whether one's work is pleasant or unpleasant is partly a function of comparisons with past work, with others' work, and with the extent of reward for such work. And whether one's rewards are good or bad is judged in light of comparisons with past rewards, others' rewards, and sacrifices made. If one reasons that one's work is onerous or one's rewards are poor, then engaging in that work or accepting those rewards can flame the anger that leads to injustice deliberations. Then demonstrations of inequity, inequality, need, or whatnot might convince self and others that injustice is present, and things need to be changed. We claim, though, that the process must proceed through emotion, and if the emotion gets allayed through the many interactional circumstances we have examined, then injustice will be ignored.

Related Approaches

Although we believe that our position about the role of emotion and social factors in instigating justice deliberations is relatively novel and unique, there have been related ideas in the literature. In this section, we will discuss two theories of justice which we believe have some interesting similarities to our view. First, we will briefly describe the way that status-value theory has been applied to distributive justice. This theory, developed by Joseph Berger and his colleagues (Berger et al., 1972; Berger et al., 1977; Berger et al., 1983) attempts to explain how expectations for rewards could vary as a function of the "status characteristics" of the recipient. Despite a somewhat different focus, this approach has some interesting similarities with our application of affect control theory to distributive justice.

In this section, we will also discuss recent work in procedural justice, which argues 1) that judgments of the fairness of a procedure for deciding outcomes have at least as much influence on judgments of distributive fairness as information about the outcome itself (e.g Lind, 1992), and 2) that decisions about the fairness of a procedure are based on social or interactional factors, not predominantly on factors directly related to the procedure itself (Lind, 1992; Tyler & Lind, forthcoming).

Status-value Theory. Status-value theory (Berger et al., 1977; Humphreys and Berger, 1981) as applied to distributive justice (Berger et al., 1972; Berger et al., 1983) argues that distributive fairness is determined by a comparison of the rewards an individual receives to the rewards that the person expects. This, basically, is the comparison postulate discussed above. However, Berger and his colleagues add an interesting dimension to this view of justice by specifying a means whereby the expected reward in a specific (local) situation can derive from the broader social context—from a referential structure that arises from considering factors outside the local situation.

The referential structure refers to the generalized ways in which "meaning is given to rewards and expectations are formed about their allocation." (Berger et al., 1972, p. 119). Status value theory attempts to explain how the cultural meanings of interactants, rewards, and expectations are translated from this referential structure to the local situation.

The referential structure has much in common with the notion of fundamental sentiments as they are conceptualized in affect control theory. In both cases, certain information about the relationships and meaning of various elements in a particular interaction are generated from some culturally-shared structure.[10] In status-value theory, the referential structure specifies the relationships between various status characteristics of an individual and the goal-objects (rewards) that should be expected by someone who possesses those status characteristics. One aspect of this process implies that the rewards to be expected are generated based on what a generalized other in a similar role position would be entitled to (see also Blalock & Wilken, 1979). The comparison standard in this situation is this generalized other.

As in affect control theory, then, the degree to which a particular situation will ultimately be considered unfair is dependent upon how the perceiver assigns the identity of an individual reward recipient. If the individual perceives the recipient (which may be self) as a mechanic, for example, then the generalized other to which he or she will compare is a generalized mechanic. However, the assignment of another identity would result in a different conclusion as to which generalized other to use as a comparison, and therefore, which standard of pay is fair. For example, if, instead of seeing oneself as a mechanic, a person sees her- or himself as a gas station attendant, then the person might view a lesser reward as fair.

Of course, the difference between our view and the approach taken by status-value theory relates to the process by which these differences in identity assignment affect justice evaluations. Berger et al., (1972; Berger et al., 1983) propose that identification influences justice by affecting the choice of a standard for comparison. We have argued that identity may also influence justice judgments because identity affects whether or not justice-related emotions will be aroused, and therefore whether or not justice issues will be considered at all.

Another similarity between status-value theory and affect control theory has to do with Berger et al.'s (1972) position regarding the way that the meaning of status creates an expectancy for the local situation. They write:

Given a specific definition of the status significance of characteristics and goal objects in the local system, and specific beliefs about what one has a right to expect, the actual association of characteristics and goal objects in the local system either coincides with expectations or it does not. If it does, it follows that (1) the status values actually associated in the local system are balanced and (2) the system is `moral' or `just,' in the sense that it behaves in the way one has a legitimate right to expect. (p. 135)

In other words, a local system is unbalanced and unjust when there are people in it who do not have the characteristics, and do not receive the rewards, that people with that status should have. This is the same as saying that a local system is "unbalanced" to the extent that the identities of the participants are not confirmed. According to affect control theory, this will produce certain emotions. However, as we have delineated above, this does not automatically lead to attributions of injustice; the emotion could be resolved by other means.

Procedural Justice. Beginning with the work of Thibaut, Walker, and their colleagues (e.g. Thibaut & Walker, 1975), there has been an effort to explain how people evaluate the fairness of the procedures that are used to arrive at decisions. There are several reasons why this is relevant to questions of distributive justice. Often, the procedures used to decide upon a distribution are known by the participants in an interaction. When this is the case, the procedures used to determine the distribution of reward is an important part of the question of when distributions are considered fair. It is a question that, until recently, has been largely ignored in the distributive justice literature. However, some research has suggested that questions of whether or not a procedure for allocating rewards is fair are more important in decisions of distributive justice than any variables related to the size of the rewards allocated (see, e.g. Lind, 1992; see Tyler & Lind, forthcoming, for a review).

In conjunction with this finding about the importance of procedural factors in the judgment of the fairness of distributions, recent work by Tom Tyler and E. Allan Lind and their colleagues (cf. Lind, 1992; Lind & Earley, forthcoming; Lind & Tyler, 1988, chapter 10, and especially Tyler & Lind, forthcoming) have argued quite convincingly that "people use impressions of authorities and inferences about their own place in the social scheme to generate a global impression of the fairness of an organization" (Lind, 1992, p. 17). According to this view, procedural justice judgments, even in formalized institutional and organizational contexts, are made on the same basis as more interpersonal impressions.

Tyler and Lind (forthcoming) have analyzed the factors that lead up to judgments that a procedure is fair. This analysis has led them to propose what they call a "relational model" of procedural justice. Procedures are judged to be fair based on the degree to which they give the participant standing, and the degree to which they generate trust in the authorities or procedures. Standing refers to the degree to which the participant feels that those who have the most importance in the procedure give the participant full status. Trust refers to the degree to which the authority or other important people in the procedure "appear to be trustworthy, benevolent, and unbiased" (Lind, 1992, p. 17). In other words, procedures are judged to be fair based on the degree of liking, trustworthiness, and status that the participants in the procedure (decision-makers and beneficiaries of the decision) are given. Affect control theory would make the same predictions about when judgments of injustice would be considered. Our reading of Tyler and Lind's work suggests that the procedures where trust and standing are higher are situations where the fundamental identity of the actors has a higher evaluation. The differences in perceptions of the actors when trust and standing are or are not present can be compared to the differences that appear when actors apply trait adjectives (e.g. trustworthy or benevolent) to identities (e.g. the benevolent judge - cf. Averett & Heise, 1987). Procedures which are high in standing and/or trust would create a situation where the identities of the participants are particularly high in evaluation. In these situations, it would be unlikely for either actor to experience anger or guilt.

This is especially true because the evaluation of identities of the participants in most of Tyler & Lind's research are already fairly high. Much of their research, for instance, was conducted in courtrooms and other legal settings (e.g. court-annexed arbitration proceedings). The authority in this case is typically a judge (or an arbitrator). Judges are generally seen as somewhat good (for example, Canadian females rate judges at 1.42 on the evaluation dimension). Combining this positive identity with a trait adjective such as "benevolent" (the benevolent judge) would raise the evaluation even higher. Thus, we see our model as consistent with the theoretical and empirical work on procedural justice.

However, the two models (our model and the "relational model") are not redundant. For instance, one prediction of our model that is not explicit in Tyler and Lind's (forthcoming) relational model of justice (but which is not necessarily inconsistent with it) is that any other factors that might influence the ascribed identities of actors in the situation, and therefore would affect the likelihood of injustice-related emotions, would also affect the likelihood that justice considerations would arise. These factors might include the setting of the decision, previous behaviors of the actors, etc. In addition, it is possible that a sequence of negative events might push the transient impressions of participants down sufficiently that even a procedure decided on by a "benevolent judge" would evoke injustice-related emotions, and therefore justice considerations.

Another contribution of our approach is that ACT can specify specific identities that have different affective meanings. That is, while Tyler & Lind suggest that procedures which raise the evaluation of participants will be viewed as fairer, our approach suggests specific types of people which will increase or decrease the likelihood that justice will be considered.

We feel, therefore, that our approach compliments that of Tyler and Lind (forthcoming). In addition to suggesting additional factors that might affect procedural justice judgements, and specific identities that might influence judgements of justice, we provide a model for how the "relational" aspects of justice judgements operate—by means of affect and emotion.

Summary and Conclusion

We have presented a theory which attempts to show how emotion can be generated in situations which typically arouse justice concerns, and we have suggested that these justice-related emotions instigate deliberations about whether or not a particular situation is fair. From this perspective, the emotions typically associated with judgments of justice are seen as being a contributing cause of a decision that a transaction is unfair, rather than being a consequence of these judgments. This approach suggests that factors affecting emotion may be important elements in the process by which decisions to demand reparations, and to attempt to reorganize the means of distribution of resources are made.

We feel that the potential impact of emotion on this process is rather dramatic. This paper has been an attempt to highlight some of these impacts. The ultimate test, of course, is empirical, and we await further research to delineate the emotional bases of injustice.


Completion of this paper was facilitated by support from a National Institutes of Mental Health Training Grant in Measurement of Affect and Affective Processes (PHS T32 MH 15789-14). We would also like to thank A. George Alder, Guillermina Jasso, Barry Markovsky, Robert Sinclair, and one anonymous reviewer for helpful comments on an earlier version of this manuscript.


Adams, J.S. (1965). Inequity in social exchange. In L. Berkowitz (Ed.) Advances in Experimental Social Psychology, (vol. 2) New York: Academic Press.

Averett, C. P., and D. R. Heise. (1987). Modified social identities: Amalgamations, Attributions, and Emotions. Journal of Mathematical Sociology, 13, 103-132.

Austin, W., N.C. McGinn, and C. Susmilch (1980). Internal standards revisited: Effects of social comparisons and expectancies on judgments of fairness and satisfaction. Journal of Experimental Social Psychology, 16, 426-441.

Berger, J., M.H. Fisek, F.Z. Norman and M. Zelditch, Jr. (1977). Status Characteristics and Social Interaction: An Expectation States Approach. New York: Elsevier.

Berger, J., M. Zelditch, Jr., B. Anderson, and B.P. Cohen (1972). Structural aspects of distributive justice: A status value formulation. In J. Berger, M. Zelditch, Jr., & B. Anderson, (Eds.). Sociological Theories in Progress. (pp. 119-146). Boston: Houghton Mifflin Company.

Berger, J., M.H. Fisek, R.Z. Norman, and D.G. Wagner (1983). The formation of reward expectations in status situations. In D.M. Messick & K.S. Cook (Eds.), Equity Theory: Psychological and Sociological Perspectives. (pp. 127-168). New York: Praeger

Blalock, H.M., Jr. and P.H. Wilken (1979). Intergroup processes: A Micro-Macro Perspective. New York: Free Press.

Deutsch, M. (1979). A critical review of 'Equity Theory': An alternative perspective on the social psychology of justice. International Journal of Group Tensions, 1, 20-49.

Deutsch, M. (1983). Current social psychological perspectives on justice. European Journal of Social Psychology, 13, 305-319.

Deutsch, M. (1985). Distributive Justice: A Social-Psychological Perspective. New Haven, Ct.: Yale University Press.

Fazio, R. and J. Cooper (1983). Arousal in the dissonance process. In J.T. Cacioppo & R.E. Petty (Eds.), Social Psychophysiology: A Sourcebook. (pp. 122-152). New York: The Guilford Press.

Fiedler, K. (1991). Emotional mood, cognitive style, and behavior regulation. In K. Fiedler and J. Forgas (Eds.), Affect, Cognition and Social Behavior: New Evidence and Integrative Attempts. (pp. 100-119) Toronto: C.J. Hogrefe.

Gallagher, D. and G.L. Clore (1985). Effects of fear and anger on judgments of risk and evaluations of blame. Paper presented at the annual meeting of the Midwestern Psychological Association, Chicago, Ill.

Goffman, E. (1963). Stigma: Notes on the Management of Social Identity. Englewood Cliffs, N.J.: Prentice Hall.

Gollob, H. F. (1968). Impression formation and word combination in sentences. Journal of Personality and Social Psychology, 10, 341–353.

Gollob, H.F., and B.B. Rossman, (1973). Judgments of an actor's 'Power and ability to influence others.' Journal of Experimental Social Psychology, 9, 391–406.

Harris, R.J. (1976). Handling negative inputs: on the plausible equity formulae. Journal of Experimental Social Psychology, 12, 194-209.

Heise, D. R. (1966). Social status, attitudes and word connotations. Sociological Inquiry, 36, 227-239.

Heise, D. R., (1969). Affective dynamics in simple sentences. Journal of Personality and Social Psychology, 11, 204–213.

Heise, D.R., (1970). Potency dynamics in simple sentences. Journal of Personality and Social Psychology, 16, 48–54.

Heise, D. R. (1979). Understanding Events: Affect and the Construction of Social Action. New York: Cambridge University Press.

Heise, D.R. (1985). Affect control theory: Respecification, estimation, and tests of the formal model. Journal of Mathematical Sociology, 11, 191-222.

Heise, D.R. (1989). Effects of emotion displays on social identification. Social Psychology Quarterly, 52, 10-21.

Heise, D.R. and L. Smith–Lovin, (1981). Impressions of goodness, powerfulness and liveliness from discerned social events. Social Psychology Quarterly, 44, 93–106.

Heise, D. and E. Lewis. (1988). Programs Interact and Attitude: Software and Documentation. Dubuque, Iowa: Wm. C. Brown Publishers, Software.

Heise, D. R., and L. Thomas (1989). Predicting impressions created by combinations of emotion and social identity. Social Psychology Quarterly, 52, 141-148.

Higgins, E.T. (1987). Self-discrepancy: A theory relating self and affect. Psychological Review, 94, 319-340.

Homans, G.C. (1961). Social Behavior: Its Elementary Forms. New York: Harcourt, Brace, and World.

Humphreys, P. and J. Berger (1981). Theoretical consequences of the status characteristics formulation. American Journal of Sociology, 86, 953-983.

Isen, A. (1987). Positive affect, cognitive processes, and social behavior. In L. Berkowitz (Ed.). Advances in Experimental Social Psychology, (Vol. 20, pp. 203-253). New York: Academic Press.

Isen, A. and K.A. Daubman (1984). The influence of affect on categorization. Journal of Personality and Social Psychology, 47, 1206-1217.

Jasso, G. (1978). On the justice of earnings: A new specification of the justice evaluation function. American Journal of Sociology, 83, 1398-1419.

Jasso, G. (1980). A new theory of distributive justice. American Sociological Review, 45, 3-32.

Jasso, G. (1983). Social consequences of the sense of distributive justice: Small-group applications. In D.M. Messick & K.S. Cook (Eds.), Equity Theory: Psychological and Sociological Perspectives, New York: Praeger.

Jasso, G. (1986). A new representation of the just term in distributive-justice theory: Its properties and operation in theoretical derivation and empirical estimation. Journal of Mathematical Sociology, 12, 251-274.

Jasso, G. (1990). Methods for the theoretical and empirical analysis of comparison processes. In C.C. Clogg (Ed.) Sociological Methodology 1990 (pp. 369-419), Wash., D.C.: American Sociological Association.

Jones, E.E., et al. (1984). Social Stigma: The Psychology of Marked Relationships. New York: W.H. Freeman.

Kemper, T. D. (1990). Social relations and emotions: A structural approach. In T. D. Kemper (Ed.), Research Agendas in the Sociology of Emotions. Albany: State University of New York Press.

Lakoff, G. (1987). Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago: The University of Chicago Press.

Leventhal, G.S., J. Karuza Jr., and W.R. Fry (1980). Beyond fairness: A theory of allocation preferences. Justice and Social Interaction: Experimental and Theoretical Contributions from Psychological Research, New York: Springer-Verlag.

Lerner, M.J. (1980). The Belief in a Just World: A Fundamental Delusion. New York: Plenum Press.

Lind, E.A. (March, 1992). The fairness heuristic: Rationality and "relationality" in procedural evaluations. Paper presented at the Fourth International Conference of the Society for the Advancement of Socio-Economics, Irvine, Ca.

Lind, E.A. and P.E. Earley (forthcoming). Procedural justice and culture. International Journal of Psychology.

Lind, E.A. and T.R. Tyler (1988). The Social Psychology of Procedural Justice. New York: Plenum Press.

MacKinnon, N. J. (forthcoming). Analyzing Social Interaction: A Cross-Cultural Study in Affect Control Theory Albany: SUNY Press.

MacKinnon, N.J. and D.R. Heise (forthcoming). Affect control theory: Delineation and development. In J. Berger and M. Zelditch, Jr. (Eds.). Theoretical Research Programs: Studies in Theory Growth. Stanford, Ca.: Stanford University Press.

Mark, M.M. and R.C. Sinclair (1992). Mood, perceived justice, and categorization breadth: A processing strategy interpretation. Unpublished Manuscript, Pennsylvania State University, University Park, Pa.

Markovsky, B. (1985). Toward a multilevel distributive justice theory. American Sociological Review, 50, 822-839.

Melton, R.J. and S. J. Scher (1992). The role of emotion in judgments of transgressors. Paper presented at the annual meetings of the Midwestern Psychological Association, Chicago, Ill.

Messick, D.M. and K. P. Sentis (1979). Fairness and preference. Journal of Experimental Social Psychology, 15, 418-434.

Messick, D.M. and K. Sentis (1983). Fairness, preference, and fairness biases. In D.M. Messick & K.S. Cook (Eds.), Equity Theory: Psychological and Sociological Perspectives, New York: Praeger.

Mikula, G. and T. Schwinger (1978). Intermember relations and reward allocation: Theoretical considerations of affects. In H. Brandstätter, J.H. Davis, and Heinz Schuler (Eds.), Dynamics of Group Decisions. Beverly Hills, Ca.: Sage Publications.

Murray, N., H. Sujan, E.R. Hirt, and M. Sujan (1990). The influence of mood on categorization. Journal of Personality and Social Psychology, 59, 411-425.

O'Malley, M.N. and D.K. Davies (1984). Equity and affect: The effects of relative performance and moods on resource allocation. Basic and Applied Social Psychology, 5, 273-282.

Osgood, Charles E. 1962. Studies of the generality of affective meaning systems. American Psychologist, 17, 10-28.

Osgood, C. H., W. H. May, and M. S. Miron. (1975). Cross–Cultural Universals of Affective Meaning. Urbana: University of Illinois Press.

Russell, J. (1991). In defense of a prototype approach to emotion concepts. Journal of Personality and Social Psychology, 60, 37-47.

Schwarz, N. (1983). Stimmung als information: Zum einfluß von stimmungen auf die beurteilung des eigenen llebens. In G. Luers (Ed.), Beright über den 33. Kongreß der Deutschen Gesellschaft für Psychologie in Mainz 1982. Göttingen: Hogrefe [Cited in Schwarz, 1990].

Schwarz, N. (1990). Feelings as information: Informational and motivational functions of affective states. In E.T. Higgins & R.M. Sorrentino (Eds.), Handbook of Motivation and Cognition: Foundations of Social Behavior (Vol. 2, pp. 527-561). New York: Guilford Press.

Schwarz, N. and G. Clore (1983). Mood, misattribution, and judgments of well-being: Informative and directive functions of affective states. Journal of Personality and Social Psychology, 45, 513-523.

Schwarz, N. and G. Clore (1988). How do I feel about it? The information function of affective states. In K. Fiedler & J. Forgas (Eds.), Affect, Cognition, and Social Behavior (pp. 44-62). Lewiston, NY: Hogrefe.

Schwarz, N., W. Servay, and M. Kumpf (1985). Attribution of arousal as a mediator of the effectiveness of fear-arousing communications. Journal of Applied Social Psychology, 15, 74-78.

Sinclair, R.C. (1988). Mood, categorization breadth, and performance appraisal: The effects of order of information acquisition and affective state on halo, accuracy, informatoin retrieval, and evaluations. Organizational Behavior and Human Decision Processes, 42, 22-46.

Sinclair, R.C. and Mark, M.M. (1991). Mood and the endorsement of egalitarian macrojustice versus equity-based microjustice principles. Personality and Social Psychology Bulletin, 17, 369-375.

Sinclair, R.C. and Mark, M.M. (1992). The influence of mood state on judgment and action: Effects on persuasion, categorization, social justice, person perception and judgmental accuracy. In L. Martin & A. Tesser (Eds.). The Construction of Social Judgments, Hillsdale, N.J.: Erlbaum.

Sinclair, R.C., Mark, M.M., Weisbrod, M.S. & Enzle, M.E. (1992). The Effect of Mood on Categoriztion Breadth: Flexibility or Differential Processing Strategy?, Unpublished Manuscript, University of Alberta, Edmonton, Alberta, Canada.

Smith-Lovin, L. (1987a). Impressions from events. Journal of Mathematical Sociology, 13, 35-70.

Smith-Lovin, L. (1987b). The affec­tive control of events within set­tings. Journal of Mathematical Sociology, 13, 71-101.

Smith-Lovin, L. & Heise, D.R. (1987). Analyzing Social Interaction: Advances in Affect Control Theory. New York: Gordon and Breach Science Publishers (Reprint of The Journal of Mathematical Sociology, 13 (1-2).)

Smith–Lovin, L., and D.R. Heise, (1988). Analyzing Social Interaction: Advances in Affect Control Theory. New York: Gordon and Breach Science Publishers. (Reprint of a special issue of the Journal of Mathematical Sociology, Vol. 13.)

Smith–Lovin, L., and D.R. Heise. (1982). A structural equation model of impression formation. In N. Hirschberg and L.G. Humphreys (Eds.), Multivariate Applications in the Social Sciences. (pp. 195–222). Hillsdale, N.J.: Lawrence Erlbaum.

Smith-Lovin, L., and W. Douglas, (forthcoming). An affect control analysis of two religious subcultures. In V. Gecas and D. Franks (Eds.), Social Perspective in Emotions, Vol. 1, JAI Press.

Stouffer, S., E. Suchman, L. DeVinney, S. Star, and R. Williams (1949). The American Soldier: Adjustment During Army Life. Princeton, N.J.: Princeton Univeristy Press.

Thibaut, J.W. and H.H. Kelley (1959). The Social Psychology of Groups. New York: Wiley.

Thibaut, J.W. and L. Walker (1975). Procedural Justice: A Psychological Analysis. Hillsdale, N.J.: Erlbaum.

Thoits, P.A. (1990). Emotional deviance: Research agendas. In T. D. Kemper (Ed.), Research Agendas in the Sociology of Emotions. Albany: State University of New York Press.

Tyler, T.R. and E.A. Lind (forthcoming). A relational model of authority in groups. In M. Zanna (Ed.). Advances in Experimental Social Psychology, (Vol. 25). New York: Academic Press.

Walster, E., E. Berscheid, and G.W. Walster (1973). New directions in equity research. Journal of Personality and Social Psychology, 25, 151-176.

Walster, E., G.W. Walster, and E. Berscheid (1978). Equity: Theory and Research. Boston: Allyn & Bacon.

Zanna, M. P. and J. Cooper (1976). Dissonance and the attribution process. In J.H. Harvey, W. Ickes, & R.F. Kidd (Eds.), New Directions in Attribution Research, (vol. 1, pp. 199-217) Hillsdale, N.J.: Lawrence Erlbaum Associates.

Zillman, D. (1978). Attribution and misattribution of excitatory reactions. In J.H. Harvey, W. Ickes, & R.F. Kidd (Eds.), New Directions in Attribution Research, (vol. 2, pp. 335-368) Hillsdale, N.J.: Lawrence Erlbaum Associates.

Zillman, D. (1983). Transfer of excitation in emotional behavior. In J.T. Cacioppo & R.E. Petty (Eds.), Social Psychophysiology: A Sourcebook (pp. 215-240). New York: The Guilford Press.


[1]. Or, at least as fair as the best available alternative (cf. Thibaut & Kelley, 1959).

[2]. This form of the function is equivilant to Jasso's because

q(RA/RB) = logJI(RA/RB),

with q = 1/[ln(JIAB)].

[3]. Recently, several authors have begun to examine when people might use other justice principles (cf. Deutsch, 1985; Mikula & Schwinger, 1978; Leventhal et al, 1980 for examples). To date, these theories have been less formal. We believe that what we say can be applied to these theories as well. However, the equity principle has been far and away the most common basis for justice research, and therefore we will not focus on these so-called "multi-principle" approaches in this paper.

[4]. Sinclair & Mark (1992) argue that mood affects justice judgements because of the effects of moods on cognitive processing strategies. In particular, people in good moods tend to process information in a simpler, more heuristically-based, non-systematic way. In contrast, people in bad moods tend to process information more carefully and systematically (cf. Fiedler, 1991; Schwarz, 1990, Sinclair & Mark, 1992 for reviews). Thus, "if positive moods lead to nonsystematic processing, failure to discriminate among stimuli, and thus the use of broader categories, then people in good moods should perceive less variation in fairness as rewards become more or less equitable, relative to subjects in other moods." (Sinclair & Mark, 1992, p. 176).

There appears to be some inconsistency between these findings and other research on categorization and mood, however. Mark & Sinclair (1992) showed the most pronounced effects for depressed subjects. Happy subjects seem only slightly different from subjects in a neutral mood in their fairness judgements. However, previous research on the effects of mood on categorization seems to suggest that only positive moods lead to categorization differences, relative to no-mood controls (e.g. Isen & Daubman, 1984; Murray et al., 1990), and if negative moods have any effect on categorization, it is in the same direction as the effect of positive mood (Isen & Daubman, 1984). Sinclair, Mark, and their colleagues (Sinclair, 1988; Sinclair and Mark, 1991, 1992; Sinclair, Mark, Weisbrod, and Enzle, 1992) have begun to provide evidence consistent with their notion that negative moods do lead to narrower or more precise categorization, and therefore, consistent with their proposed explanation for the effects of mood on justice judgements. This issue deserves further attention. Nonetheless, it is clear that negative moods do increase the likelihood that inequitable distributions will be seen as unfair, relative to equitable distributions.

[5]. This effect is asymmetrical. For subjects who had been called on a sunny day, calling attention to the weather had no effect on their reported well-being. However, since we are concerned predominantly with the negative emotions that arouse injustice deliberations, this asymmetry does not alter our conclusions.

[6]. See Fazio and Cooper, 1983; Zanna and Cooper, 1976; and Zillman 1978, 1983 for further discussion of the theoretical and empirical background of this misattribution paradigm.

[7]. This is not an exhaustive presentation of affect control theory. Interested readers can consult other sources (Heise, 1979, 1985; Smith-Lovin and Heise, 1987) for a complete specification of the theory.

[8]. The coefficients in these equations are unstandardized, in order to allow prediction of the actual values of the transient impressions. All EPA ratings were scored using a metric ranging from -4 to +4, with an assumed interval scale (cf. Smith-Lovin, 1987, p. 43 for details).

[9]. All of the generative aspects of affect control theory—predictions of behavior, of emotions, and of identity and trait attributions—are obtained by manipulating empirically-based equations mathematically under the constraints of theoretical assumptions. We refer the reader to Heise (1987, 1989) for the actual derivations.

[10]. Theoretically, affect control theory could be applied using individualized fundamental sentiments. However, culturally-shared meaning would be a major component of these sentiments. In applications of the theory, the initial sentiments are always taken to be the culturally shared, out-of-context sentiments (i.e. the mean EPA ratings).