Thomas Scheff is well-known in sociology for several things. He originated the idea of mental illness as “residual deviance,” and, in the 1970's, was engaged in a high profile debate with Walter Gove on the issue of whether labeling actually causes deviance. Today we will review his ideas on emotion.

Scheff examines the role of shame in the socialization of emotions, dismisses the idea that it is rare, and argues that it is actually the primary social emotion. This is because, as Darwin notes about blushing, it involves us thinking what others think of us, perceptions of evaluation of the self whether positive or negative. Scheff thinks that Darwin's argument can be restated as 2 propositions: (1) blushing is caused by shame; (2) shame is caused by the perception of negative evaluations of the self. Essentially, shame is a social emotion arising from the monitoring of one's own actions by viewing oneself from the standpoint of others.

That shame is a crucial emotion for adults was also prominent in the work of early psychologist William McDougall. He thought of it as one of the self-regarding sentiments, perhaps the most important one in influencing social behavior. Like Darwin, he felt that it arose out of self-monitoring and had a biological basis. Yet, he was aware that in humans shame is considerably more complex than among animals due to our elaborately developed self­consciousness.

Low Visibility Pride and Shame:

C.H. Cooley, an early interactionist, considered pride and shame to be the crucial self­feelings, in his words “the mainspring of endeavour and a chief interest of the imagination throughout life.” He also notes how a person can go along fairly satisfied without a care for others' views, but, when suddenly others turn on him in contempt, he will “perceive from the shock of being outcast and helpless that he was living in the minds of others without knowing it, just as we daily walk the solid ground without thinking how it bears us up.” Scheff feels that pride and shame a clearly implied in this view, especially the almost continuous presence of low visibility pride.

When we turn further to Cooley's famous concept of the “looking glass self,” we can see that he viewed self-monitoring in terms of 3 steps: (1) the imagination of our appearance to another person; (2) the imagination of his judgement of that appearance; and (3) some sort of self-feeling, such as pride or mortification. Note that Cooley restricts self-feeling to the 2 he considers the most significant: pride and shame. Scheff notes that what is oddly unappreciated about this famous idea is that Cooley is implying that society rests upon a foundation of pride and shame. His analysis of the social nature of the self can be summarized in terms of 2


  1. In adults, social monitoring of the self is virtually continuous, even in solitude (i.e. we are “living in the minds of others without knowing it”)
  2. Social monitoring always has an evaluative component, and gives rise, therefore, to pride and shame.

When taken together, these 2 propositions suggest a puzzle. If social monitoring of self is almost continuous, and gives rise to pride and shame, why is it that we see so few manifestations of either emotion in adult life? One possible solution would be that both are always present, but with such a low visibility that we don't notice. This gives rise to a 3rd proposition:

  1. Adults are virtually always in a state of either pride or shame, usually of a quite unostentatious kind.

Scheff says that this idea helps us better understand and operationalize self-esteem, which could be a summary measure representing the balance of pride and shame states in a person's life. This would take into account not only the duration but the intensity of these factors. This definition implies that self-esteem involves not only what one thinks, but also how one feels about oneself - primitive concerns that seem to precede thoughts, perceptions and behaviors. Perhaps because self-feelings are usually denied and disguised, they have been omitted from most scholarly discussions of self-esteem.

The Recursiveness of Unacknowledged Shame:

Scheff argues that in modern societies adults seem to be uncomfortable about manifesting either pride or shame. The very emotions of shame and pride often seem themselves to arouse shame (even blushing). As such, he argues that shame may be recursive, acting back on itself. If shame is evoked but not acknowledged, the possibility arises that one may react emotionally to one's initial emotional reaction, then react again to the second reaction, and so on.

For example, one might be ashamed of being ashamed, creating a shame-shame spiral, or angry because one is ashamed, then ashamed because one is angry, creating a shame-anger spiral. In Scheff's example:

If I am (1) humiliated by my father's innuendo, then (2) angry at my father, then

  • ashamed of my anger (“What a monster I am! Angry at my own father when he is only trying to help me!”), the next step might be (4) to be angry at myself, and (5) to be ashamed of myself because I was so upset about what I might think about as “nothing.” Even if each link in this chain lasts only a few seconds, several thousand such links would add up to hours of what might be experienced as “helpless anger,” a mild form of the shame-anger chain reaction, or, in its more intense form, “humiliated fury.”

Scheff feels that the recursive potential of unacknowledged shame may give rise to a chain reaction with no natural limit on its duration or intensity. Like the immune reaction, which is usually life enhancing, but may, under certain circumstances, run away with itself and cause damage to the body, emotions that arise in the shame context can also result in endless destruction.

Up to this point we have concentrated on the inner recursion of unacknowledged shame. However, in a social context things get much more complicated and intense. Emotions can spiral not only within, but also between parties in social interaction. What Scheff calls a “triple spiral” forms a cycle complete in 6 steps:

  • A insults B who
  • is ashamed, then
  • is almost instantaneously angry
  • B insults A, who
  • is ashamed, then
  • is angry.

At this point the cycle can begin again. One can think of this chain as 3 spirals: one within A, one within B, and one between them.

This shame-anger spiral can be completed so quickly as to be almost invisible unless one has a way of slowing things down (e.g. Retzinger's videotapes of marital arguments on slow motion). By repeatedly observing the verbal and non-verbal gestures, it is possible to, for example, demonstrate 4 full shame-anger cycles in the first 34 seconds. When observed in this way, an ordinary conversation phrased with seemingly polite words can be seen to be packed with hostile innuendo, mutual insult and revenge.

If the 2 parties are individuals, the conflict can last as long as a lifetime; mutual hatred would seem to be the intermittent replaying of triple spirals of shame and anger. If the two parties are groups, the sequence can last longer than a lifetime, since the antagonism can be transmitted from generation to generation. The recursive potential of unacknowledged shame may provide an explanation why humans alone among living creatures have an unlimited potential for destructiveness.

Low Visibility Shame:

Scheff argues next that, due to the recursive character of shame, we would expect that pride and shame would have low visibility. Even if they were widely prevalent, persons who were proud or ashamed might be ashamed of their state, and therefore hide it from others and from themselves. But if pride and shame are usually hidden, how can we study them?

Scheff notes that methods for detecting low-visibility shame were developed by

Gottschalk and Gleser (1969) and Lewis (1971). The first only deals with verbal texts, which are crudely divided into 5 categories:

  1. Shame, embarrassment
  2. Humiliation
  3. Ridicule
  4. Inadequacy
  5. Overexposure of deficiencies/private details

While few of the sentences explicitly mention shame, most illustrate a perception of negative evaluation of self. Yet the absence of theory and direct validation makes this method seem arbitrary.

The work of Lewis treats the issue of shame markers more explicitly, and connects theory, method, and concrete episodes of behavior. She conducted a systematic analysis of shame content in complete episodes of real social interaction, entire clinical sessions. Her laborious word for word analysis led her to the discovery of what she called “unacknowledged shame,”the low-visibility shame implied by Cooley. She showed that in the hundreds of clinical sessions she examined, most shame episodes were virtually invisible to the participants, and were acknowledged by neither the patient nor the therapist. She divided episodes of unacknowledged shame into 2 basic types:(1) overt, undifferentiated shame, and (2) bypassed shame.

Overt, undifferentiated shame involves painful feelings that are not identified as shame by the person experiencing them. Rather, they are labeled with a wide variety of terms that serve to disguise the shame experience (e.g. feeling foolish, stupid, ridiculous, etc.) Culture provides many of these codewords, and many of these project the inner feeling of shame onto the outside world (e.g. it was an “awkward moment”). It was not I who felt shame, but the situation that was awkward. Our very language betrays us betrays us into denying shame.

Lewis classified such terms as shame markers, occurring in a certain context, and in association with specific types of nonverbal markers. The context always involved perception of self as negatively evaluated, either by self or others. There was always a change in the patients' manner, characterized by speech disruption (e.g. stammering), lowered or averted gaze, blushing, and a sharp drop in the loudness of speech.

Both the verbal and nonverbal markers of overt shame can be characterized as forms of hiding behavior: the words hide shame under a disguising label; the nonverbal forms suggest physical hiding. Scheff found these in a study of the “moment of truth” in videotapes of Candid Camera (e.g. covering one's face). Such behaviors were also noted in Edelman's (1987) study of embarrassment in 5 European countries.

Thus, overt, undifferentiated shame occurs when:

  1. One feels the self negatively evaluated, either by self or other;
  2. Manifests “hiding” behavior, and/or
  3. Labels or associates the painful feeling with undifferentiated terms.

In these instances, the negative evaluation of self appears to cause so much pain that it interferes with the rapid production of thought and/or speech, but the pain is mislabeled.

Lewis' second category, bypassed shame, always begins with a perception of the negative evaluation of self. However, unlike the markers of overt, undifferentiated shame which are often flagrant, those of bypassed shame may be subtle and covert. Although thought and speech are not obviously disrupted, they take on a speeded-up but repetitive quality that Lewis refers to as “obsessive.”

Typically, for example, patients repeated a story or a series of stories, talking rapidly and fluently, but not quite to the point. They appear to be unable to make decisions due to insoluble dilemmas, or complained of endless internal replaying of a scene in which they felt criticized or in error. Their minds seemed so involved with the unresolved scene that they felt inadequately involved in the present, even though there was no obvious disruption. They were distracted.

These two patterns of shame appear to involve opposite styles of response. In overt shame the victim feels emotional pain to the point that it obviously disrupts thought and speech. In bypassed shame, the victim avoids the pain before it can be completely experienced, through rapid thought, speech or actions.

Scheff feels that these two types of shame correspond to his earlier distinction between over and under distanced emotion. Overt, undifferentiated shame is under-distanced, since the intense pain of embarrassment is experienced. Bypassed shame is overdistanced: one avoids the pain by stepping outside the self as if the pain were not occurring.

These two types of shame were also anticipated by Adler's discussion of how children who do not find love available at crucial points can either develop an inferiority complex, or compensate by seeking power through incessant speech, thought and actions.

Yet while Adler's is abstract and untestable, Lewis' formulation provides the foundation for a testable theory. It describes or at least strongly implies observable markers for the major concepts in the causal chain that connects them. Her work points to the events in social interaction that can be characterized as evidence as evidence of either a secure, threatened or severed attachment. By describing the observable events in shame-anger sequences, she also specifies the behavioral manifestations of Adler's structures to the point that their presence or absence can be detected in actual episodes of social interaction.

Although overt and bypassed shame present a very different appearance, they mask an underlying similarity: they are disruptive, involve rigid and distorted reactions to reality. They are equally invisible, since one is misnamed and the other ignored. These 2 basic patterns explain the puzzle of how shame might be ubiquitous, yet usually escape notice.

Lewis' analysis of shame also converges with Freud's early work on repression causing hysteria, as well as with Tompkins' work on the therapeutic expression of forgotten grief. Lewis extended the concept of repression both theoretically and empirically, laying the groundwork for her shame constructs, the description of the context and markers for unacknowledged shame, and its role in the genesis and maintenance of compulsive behavior. Most of the shame episodes she reported appeared to be not only out of awareness, but not available to it - as suggested at the outset.

Shame in Other Theories of Emotion:

In order to relate to the shame construct outlined here, it is necessary to review earlier analyses by Tompkins, Goffman and Kemper. All had important things to say about shame, but none assigned it a central role in human behavior.

Tompkins (1971) referred to shame as “humiliation,” and offered a highly specialized psychological analysis dealing almost entirely with the internal, psychological side of shame. He saw all emotion, including shame, as arising out of sudden changes, and related to the incomplete reduction of interest or joy. Yet his reasoning tends to be circular: if shame is activated by the incomplete reduction of interest or joy, what caused this incomplete reduction, and what activated the interest or joy? While describing many types of situations that may give rise to shame, he is unable to show interactions between these types because he doesn't really focus on the social contexts that produce shame. Instead he focuses on internal, physical criteria such as “stimulation density and neural firing.”

It is ironic that Tompkins places such importance on shame, yet minimizes its connection with social events. The dominant paradigm in behavioral science exactly reverse his emphasis (though downplays this). The behaviorist approach contains the social link, reinforcement through punishment and reward in the social environment (especially the classic Asch experiment). While pride and shame are not mentioned here - and even downplayed - Asch's findings of 75% yielding to majority judgement can be read in this fashion.

Erving Goffman came closest to describing a model of the interaction between deference and emotion in his analysis of interaction ritual: protection of “face” is seen as the dominant motive in social life, interactants are exquisitely sensitive to the exact amount of deference extended and received, and avoiding embarrassment is the goal that drives “face work.” Yet Goffman stopped short of a model of the interaction between inner and outer events. He stated explicitly that he was to deal only with the outer, social aspects of interaction, avoiding the inner ones, restricting his purview in order to provide a truly sociological analysis.

Kemper, finally, included social, psychological and biological concepts. He ultimately placed the causes of emotional arousal in the social structural elements of power and status. He refers to pride as important in its own right, shame as an ingredient in punishment, and calls attention to shame-anger sequences. Yet, again, Kemper's framework falls short of a comprehensive model of the role of shame in behavior. Like the behaviorists he recognizes the social link, but slights pride and shame as reinforcers. His treatments of pride and shame are essentially in passing, included in long lists of other emotions. These are incidental to his main analysis - the effect of social structural variables on emotions.

Both Tompkins and Kemper attempt to construct formal theories, but these are so abstract that they are unable to specify the major causal sequences for a model of emotions (Kemper would disagree). Tompkins reifies internal processes; Kemper social structure. Each decided that one element is causal, and minimized the other.

These a priori restrictions have the same ultimate effect: denying the role of pride and shame as causal agents. The causes they emphasize seem more objective to them, perhaps because pride and shame do not seem real in a shame-denying culture. Indeed, most formal theories are biased towards rational or material models of causation, because emotion and mood do not seem real in our civilization. Yet mood may dominate many kinds of social events and phenomena (e.g. can we see shame/anger spirals in mobs, crowds, therapy sessions, mental illness, crime, physical and verbal violence, conflict between groups and nations). Hence, the propensity to reify the kinds of causes that seem real and dignified, and to ignore those that do not, may lie at the root of our failure to solve the most fundamental problems facing us.

A Research Agenda: Socialization of Emotions and Learning:

Scheff argues that one implication of his argument is that shame, though disguised and denied, may nevertheless be crucial in the socialization of emotions. If emotions such as grief, fear or anger are aroused in a context in a context that gives rise to shame about them, then resolution of these feeling states may fail to take place. Instead, shame may serve to inhibit resolution, and, under some conditions, give rise to intense, or lengthy, chain reactions. Hence, shame in the caretaker in response to childrens' emotions could be of immense significance. Children socialized through shame may view their emotional arousal as having the potential of becoming an infinitely painful experience.

Parental shame responses to childrens' emotions can be flagrant and overt. Moreover, children may be predisposed to be extremely sensitive even to mild or disguised reactions in the caretaker. A caretaker may inadvertently pass on his or her own unrecognized patterns of shame. Perhaps disruptions of psychosexual functions could be explained in this way.

There is also evidence that mens' and womens' emotions are socialized somewhat differently. Women are usually more field dependent (prone to overt shame), and men usually more field independent (prone to bypassed shame). Perhaps field dependence is caused by socializing techniques involving overt shame, and field independence bypassed shame (e.g. silence and withdrawal). Perhaps a careful analysis of the occurrence of shame and embarrassment in the socialization of children could explain characteristic personality problems in both male and female development, as well as common difficulties in adult male-female relationships. The shame construct could also be used to revive the interest in childrearing and how it impacts on child development. The crucial distinction would not be between physical and emotional punishment, but on the manner that accompanied punishment (e.g. shaming vs. non­shaming would have a different impact on resentment toward the parent).

The relative absence of shaming in childhood socialization might also be useful in studies of the acquisition of language and other skills: how quickly and easily children learn their native tongue vs mathematics. Both are complex, but one is taught by self-confident parents who positively reinforce; the other by a relative stranger, probably not an expert, who may feel incompetent, nervous, and the contagion of shame may come to be widely prevalent.

Study Designs:

Scheff is interested in how videotape recording has been used successfully to study the moment by moment interactions between actors. He proposed building on this methodological format as a means to test hypotheses concerning the role of excessive shaming in producing compulsive behavior. Specifically, he proposes a comparison between contexts in which excessive shaming may be frequent and those in which it may be infrequent.

Scheff's idea is to look at the socialization of language in infants and very young children. He wants to look at 2 different kinds of family settings: (1) in which the frequency of intense shaming and embarrassment is low; and (2) one in which it is high. After reviewing some of behaviorist B.F. Skinner's principles that lead to accelerated learning (e.g. clear goals, little pressure, positive reinforcement, and avoiding punishment), Scheff hypothesizes that children learn complex languages due to a moratorium on shaming children in this regard (often parents react with pride even to a child's mistakes). Essentially, Scheff thinks that such a study designed to contrast socialization in language, where shaming is expected to be rare, with areas where shame and embarrassment are prevalent, might bring socialization practices and their results into high relief.

Scheff argues that 4 areas where high levels of shame and embarrassment might be expected are (a) the socialization of the infant's cry; (b) its body functions; (c) genital touching; and (d) disputes. The recording of infant-caretaker interaction in the first 3 areas could begin very early in the infant's life. The arousal of shame and embarrassment in the caretaker might be of low intensity, requiring careful analysis of facial expression and other non-verbal markers.

Because shame arousal in the caretaker might be subtle, the comparison between its expressions in these areas with those that occur in language acquisition would be very important. One could expect that infants would be sensitive to both the presence and absence of emotional arousal in caretakers.

A second study would also explore the possibility that rigid patterns for socializing emotions are transmitted from generation to generation. It would compare dispute tactics within and between 3 different generations, requiring a moment by moment analysis of disputes between parents and their children, between the parents, and between parents and grandparents. If similar dispute tactics are found to characterize conflict at all 3 levels, the hypothesis of social transmission would be supported.

As noted, videotaping would be the preferred method for collecting such data, adjusted according to the situations under study (e.g. asking people to discuss matters of frequent argument; asking parents to videotape interactions with children in which the parent reports the most conflict, such as going to bed).

The socialization model outlined here suggests that some parents would socialize the child's emotions during disputes with virtually the same tactics that were used by grandparents in similar situations when they were young. A child's expression of anger might be met with parental anger or silent disapproval. The former might give rise to the child developing a style of over shame about feelings of anger; the second bypassed shame in this regard. Since such tactics are compulsive and ordinarily outside awareness, the probability that the child or the parent's dispute tactics would change would be low.

However, other parents might be more tolerant of the child's anger. Parents whose anger was tolerated by their parents might also tolerate their own children's anger - as signs of uniqueness or growth. If the theory outlined here provides an accurate account of socialization, then it would explain both effective and disruptive learning with a single model. The agenda outlined here might carry forward the current interest in emotions into many new areas of research.