Book Search:  


Google full text of our books:


A Micro-sociological Theory
Randall Collins

Book Description | Endorsements | Table of Contents

COPYRIGHT NOTICE: Published by Princeton University Press and copyrighted, © 2007, by Princeton University Press. All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher, except for reading and browsing via the World Wide Web. Users are not permitted to mount this file on any network servers. Follow links for Class Use and other Permissions. For more information, send e-mail to

This file is also available in Adobe Acrobat PDF format


The Micro-sociology of Violent Confrontations

THERE IS A VAST ARRAY of types of violence. It is short and episodic as a slap in the face; or massive and organized as a war. It can be passionate and angry as a quarrel; or callous and impersonal as the bureaucratic administration of gas chambers. It is happy as drunken carousing, fearful as soldiers in combat, vicious as a torturer. It can be furtive and hidden as a rape-murder, or public as a ritual execution. It is programmed entertainment in the form of sporting contests, the plot tension of drama, the action of action-adventure, the staple shocker of the news edition. It is horrible and heroic, disgusting and exciting, the most condemned and glorified of human acts.

This vast array can be explained by a relatively compact theory. A few main processes, in combination and in differing degrees of intensity, give the conditions for when and how the various forms of violence occur.

Two moves will set up the analysis. First, put the interaction in the center of the analysis, not the individual, the social background, the culture, or even the motivation: that is to say, look for the characteristics of violent situations. That means looking for data that gets us as close as possible into the dynamics of situations. Second, compare across different kinds of violence. We need to break down the usual categories—homicides in one research specialty, war in another, child abuse in another, police violence yet elsewhere—and look for the situations that occur within them. Not that all situations are the same; we want to compare the range of variation in situations, which affects the kind and amount of violence that emerges. This will turn the wide variety of violence into a methodological advantage, giving clues to the circumstances that explain when and in what manner violence unfolds.


Not violent individuals, but violent situations—this is what a micro-sociological theory is about. We seek the contours of situations, which shape the emotions and acts of the individuals who step inside them. It is a false lead to look for types of violent individuals, constant across situations. A huge amount of research has not yielded very strong results here. Young men, yes, are most likely to be perpetrators of many kinds of violence. But not all young men are violent. And middle-aged men, children, and women are violent too, in the appropriate situations. Similarly with background variables such as poverty, race, and origins in divorce or single-parent families. Though there are some statistical correlations between these variables and certain kinds of violence, these fall short of predicting most violence in at least three aspects:

First, most young men, poor people, black people, or children of divorce do not become murderers, rapists, batterers, or armed robbers; and there are a certain number of affluent persons, white people, or products of conventional families who do. Similarly, the much asserted explanation that violent offenders are typically past victims of child abuse accounts for only a minority of the cases.1

Second, such analysis conveys a plausible picture of the etiology of violence only because it restricts the dependent variable to particular categories of illegal or highly stigmatized violence; it does not hold up well when we broaden out to all kinds of violence. Poverty, family strain, child abuse, and the like do not account for police violence or for which soldiers do the most killing in combat, for who runs gas chambers or commits ethnic cleansing. No one has shown that being abused as a child is likely to make someone a cowboy cop, a carousing drunk, or a decorated war hero. No doubt there are readers who will bridle at the suggestion; for them, violence naturally falls into hermetically sealed sections, and “bad” social conditions should be responsible for “bad” violence, whereas “good” violence—which is not seen as violence at all, when it is carried out by authorized state agents—is not subject to analysis since it is part of normal social order. In this way of thinking, there is an intermediate category of innocuous or “naughty” violence (i.e., carousing that gets out of hand), or violence that is committed by “good” persons; this is explained, or explained away, by another set of moral categories. Such distinctions are a good example of conventional social categories getting in the way of sociological analysis. If we zero in on the situation of interaction—the angry boyfriend with the crying baby, the armed robber squeezing the trigger on the holdup victim, the cop beating up the suspect—we can see patterns of confrontation, tension, and emotional flow, which are at the heart of the situation where violence is carried out. This is another way of seeing that the background conditions—poverty, race, childhood experiences—are a long way from what is crucial to the dynamics of the violent situation.

Third, even those persons who are violent, are violent only a small part of the time. Consider what we mean when we say that a person is violent, or “very violent.” We have in mind someone who is a convicted murderer, or has committed a string of murders; who has been in many fights, slashed people with a knife, or battered them with fists. But if we consider that everyday life unfolds in a chain of situations, minute by minute, most of the time there is very little violence. This is apparent from ethnographic observations, even in statistically very violent neighborhoods. A homicide rate of ten deaths per 100,000 persons (the rate in the United States peaking in 1990) is a fairly high rate, but it means that 99,990 out of 100,000 persons do not get murdered in a year; and 97,000 of them (again, taking the peak rate) are not assaulted even in minor incidents. And these violent incidents are spread out over a year; the chances of murder or assault happening to a particular person at any particular moment on a particular day during that year are very small. This applies even to those persons who actually do commit one or more murders, assaults, armed robberies, or rapes (or for that matter, cops who beat up suspects) during the course of the year. Even those persons who statistically commit a lot of crime scarcely do so at a rate of more than once a week or so; the most notorious massacres in schools, workplaces, or public places, carried out by lone individuals, have killed as many as twenty-five persons, but generally within a single episodes (Hickey 2002; Newman et al. 2004). The most sustained violent persons are serial killers, who average between six and thirteen victims over a period of years; but these are extremely rare (about one victim per five million population), and even these repeat killers go months between killings, waiting for just the right situation to strike (Hickey 2002: 12–13, 241–42). Another kind of rare cluster of violence, crime sprees, may continue for a period of days, in a chain of events linked closely by emotions and circumstances so as to comprise a tunnel of violence. Leaving these extended sequences of violence aside for the moment, I want to underline the conclusion: even people that we think of as very violent—because they have been violent in more than one situation, or spectacularly violent on some occasion—are violent only in very particular situations.2 Even the toughest hoodlums are off duty some of the time. Most of the time, the most dangerous, most violent persons are not doing anything violent. Even for these people, the dynamics of situations are crucial in explaining what violence they actually do.


Surveys of individuals orient our theories to the characteristics of individuals, packaged in the terms of standard sociological variables. To move to a sociological theory, not of violent individuals, but of violent situations, we must emphasize a different way of collecting and analyzing data. We need direct observation of violent interaction to capture the process of violence as it actually is performed. Our theories are constrained by having been based upon statistics assembled after the fact, packaged by the criminal justice system, or upon interviews with convicted prisoners or other participants. Victim surveys are a step in the right direction, but they remain limited, not only by the issue of to what extent victims are telling the truth, but also by the problem that persons are generally not good observers of the details and contexts of dramatic events. Our ordinary discourse does not provide the language in which to describe micro-interaction well; instead, it offers a set of clichés and myths that predetermine what people will say. This is true also of military violence, riots, sports violence, or even ordinary quarrels; when participants talk about violent situations, they tend to give a very truncated, and by their own lights, idealized version of what went on.

A new era has emerged in recent decades as it has become possible to study violence as recorded on video tape from security systems, police recordings, and news and amateur video photographers. When ordinary observers see such recordings, they are usually shocked. A riot eventually followed the publicity given to a video recording, taken by an amateur with a new camcorder, of the Rodney King arrest in Los Angeles in 1991. Events are always interpreted in terms of prevailing ideological categories; the concepts easily at hand were those of a racially motivated beating. But what was so shocking about the Rodney King video was not its racial aspect; it was the beating itself, which did not look at all like what we think violence is supposed to look like. Visual evidence shows us something about violence that we are not prepared to see. The pattern looks much the same in a wide range of incidents, in many different ethnic combinations within and across ethnic group lines (we will examine some of these in chapters 2 and 3). Racism may contribute to building up some situations of violence, but it is one lead-in condition among others, and neither a necessary nor a sufficient condition; the situation of violence itself has a dynamics that is more pervasive than racism.

Violence as it actually becomes visible in real-life situations is about the intertwining of human emotions of fear, anger, and excitement, in ways that run right against the conventional morality of normal situations. It is just this shocking and unexpected quality of violence, as it actually appears in the cold eye of the camera, that gives a clue to the emotional dynamics at the center of a micro-situational theory of violence.

We live in an era in which our ability to see what happens in real-life situations is far greater than ever before. We owe this new vision to a combination of technology and sociological method. The ethnomethodologists of the 1960s and 1970s took off as an intellectual movement in tandem with the use of newly portable cassette tape recorders; this made it possible to record at least the audio part of real-life social interactions, and to play it back repeatedly, slowing it down and subjecting it to analysis in a way that had been barely possible with fleeting observations in real time, giving rise to the field of conversation analysis (Sacks, Schegloff, and Jefferson 1974; Schegloff 1992). As video recording devices became more portable and ubiquitous, it has been possible to look at other aspects of micro-behavior, including bodily rhythms, postures, and expressions of emotion. Thus it is not surprising that the period from about 1980 onward has been the golden age for the sociology of emotions (Katz 1999, among many others).

It is not literally true that a picture is worth a thousand words. Most people will not see what is in a picture, or will see it through the most readily available visual clichés. It takes training and an analytical vocabulary to talk about what is in a picture, and to know what to look for. A picture is worth a thousand words only for those who already have internalized an adequate vocabulary. This is particularly so when we have to train ourselves to see micro-details: the movements of some facial muscles rather than others that distinguish a false smile from a spontaneous one; the movements that display fear, tension, and other emotions; the smoothness of rhythmic coordination and the hitches that indicate disattunement and conflict; the patterns in which one person or another seizes the initiative and imposes a rhythm upon others. The methods of visual and auditory recording now available open up the potential to see a vast new landscape of human interaction; but our ability to see goes in tandem with the expansion of our theories of what processes are out there to be seen.

This is so also in the micro-sociology of violence. The video revolution has made available much more information about what happens in violent situations than ever before. But real-life recording conditions are not like Hollywood film studios; lighting and composition are far from ideal, and the camera angles and distance may not be just the ones a micro-sociologist would prefer. We need to disengage ourselves from the conventions of dramatically satisfying film (including TV commercials) where the camera cuts to a new angle every few seconds at the most, and a great deal of editing has gone on to juxtapose an interesting and engaging sequence. A micro-sociologist can spot the difference between raw observational recording and artistically or editorially processed film, usually within seconds. Raw conflict is not very engaging, for all sorts of reasons; as micro-sociologists, we are not in it for entertainment.

Other approaches besides live video have opened up the landscape of violence as it really happens. Still photography has gotten better throughout the past century and a half; cameras have become more portable, and lenses and lighting devices have made it possible to capture scenes that previously would have been limited to static posed shots in relatively sheltered conditions. Professional photographers have become more intrepid, particularly in riots, demonstrations, and war zones; the number of photographers killed has gone up drastically in the past ten years, far above any previous period.3 This too is an opportunity for micro-sociologists, although the aforementioned caveats again apply. Still photos are often better than videos for capturing the emotional aspects of violent interaction. When we analyze a video of a conflict sequence (or indeed any video of interaction), we may slow it down to segments of micro-seconds (frame-by-frame in older camera film) to pull out just those details of bodily posture, facial expression, and sequence of micro-movements. In depictions of riots, which I use extensively in this work, still photos dramatically show the division between the active few on the violent front and the supporting mass of demonstrators. The danger is in assuming one can read the still photo without sociological sensibilities. Highly artistic or ideological photographers are less useful here than routine news photographers; some photos of demonstrations or combat have an artistic or political message that governs the whole composition; we need to look from a different vantage point to get at the micro-sociological aspects of conflict.

An intellectual stance on what to look for has gone along with technological advances, and sometimes preceded them. The military historian John Keegan (1976) set out to reconstruct battles from the ground up, investigating what must actually have happened as each segment of troops rushed forward or fell down; as horses, men, and vehicles got tangled in traffic jams; as weapons were wielded skillfully, accidentally, or not at all. Other military analysts have found out how many guns were loaded when recovered from dead troops on battlefields; and historical battles have been reconstructed with laser beams. What we have learned about soldiers in combat has opened the door for understanding violent situations in general. The emotional relationships between soldiers and their comrades, and between them and their equally human enemies, provided one of the first clues to how violent situations unfold.4

In our ordinary compartmentalized way of viewing things, it is a leap from military history to reconstructions of police violence, but the methodological and theoretical parallels are strong. We can understand the occasions on which police are violent by techniques such as video recordings and through methods of reconstructing events, such as ballistics analysis of the trajectories of bullets, how many hit intended and unintended targets, and how many missed entirely. Old-fashioned ethnographies have helped too; ride-alongs by sociologists in police cars, dating to the 1960s, preceded some of these technological advances and provided some key theoretical components. Technologies by themselves rarely provide real insight; it is their combination with analytical viewpoints that is crucial.

To summarize, there are at least three methods for getting at situational details of violent interactions: recordings, reconstructions, and observations. They are most useful when used in combination.

Technologies of recording real-life conflict are useful for a series of reasons: they can provide us details that we otherwise wouldn’t see at all, that we were not prepared to look at, or did not know were there; they can give us a more analytical stance, more detached from the everyday perceptual gestalts and the clichés of conventional language for talking about violence; they enable us to look at a situation over and over, getting beyond the initial shock (or jadedness, prurient interest, and the like) so that we can bring our theoretical minds to bear, and to make discoveries or test theories.

Reconstructions are important because violent situations are relatively rare, and for many incidents we would most like to understand, there were no recording devices available at the time. We are not as much in the dark as we once thought we were: as we have gotten better at situational analysis, and (coming from another angle) as new techniques keep on being developed for analyzing physical clues left on the ground, it has been possible to reconstruct many violent scenes. A wide range of reconstructions is useful to us, including historical events, because they give us theoretical leverage for finding both the commonalities and the dimensions of variation among violent situations.

Finally, there is human observation. This can be old-fashioned ethnography, especially the participant observation version in which the sociologist (or anthropologist, psychologist, or sophisticated journalist) gets inside the scene with his or her senses sharpened, looking for telling details. A variant is equally old-fashioned self-observation, reports on what oneself experienced as a participant. In the field of violence, much of what we have learned comes from reports by ex-soldiers, ex-criminals, or indeed not so “ex” persons, who are reflective enough to talk about fights they have seen or been in. There is also much of value here in reports from victims of violence, although this has not been much exploited by sociologists, beyond bare statistical counts of how often certain kinds of victimization occurs. Moreover, as we gain a better theoretical understanding of what are the important micro-details of violent confrontations, we become better at interrogating our own experience, and better at asking retrospective observers for the kind of details we would like to know about their encounters with violence. By providing a vocabulary, we make our informants often quite good reporters of details they otherwise gloss over.

The three kinds of situational evidence fit together. They complement each other not only ethodologically but also substantively. They all reveal a common situational dynamic. That is what this book is about.


To develop a theory of the dynamics of violence requires another shift: to work across research specialties, rather than be confined within them. The center of this approach is to compare different kinds of violence in a common theoretical framework. Is this not to compare apples and oranges, or at best merely to taxonomize? This is a point that cannot be decided a priori. Once we look, we find that violence is an array of processes that all follow from a common situational feature of violent confrontations.

I will state the point cryptically here: violence is a set of pathways around confrontational tension and fear. Despite their bluster, and even in situations of apparently uncontrollable anger, people are tense and often fearful in the immediate threat of violence—including their own violence; this is the emotional dynamic that determines what they will do if fighting actually breaks out. Whether indeed that will happen depends on a series of conditions or turning-points that shape the tension and fear in particular directions, reorganizing the emotions as an interactional process involving everyone present: the antagonists, audience, and even ostensibly disengaged bystanders.

How do we know this? The theoretical point has developed from accumulating information on a variety of violent situations. The first breakthrough came from the study of military combat. Fear, wild firing, hitting soldiers on one’s own side, freezing up: these were features noted by officers analyzing the behavior of frontline troops in battle, beginning with the nineteenth-century French officer Ardant du Picq, who collected questionnaires from combat officers. S.L.A. Marshall got closer to the immediate action in his World War II post-battle interviews with soldiers themselves. In the 1970s, the picture of battle behavior was systematized in historical reconstructions by Keegan and others. By the 1990s, the military psychologist Dave Grossman synthesized a theory of combat centered on the management of fear. An even more pronounced pattern of alternating fearful and aggressive behavior is seen in ethnographic films made in the 1960s of fighting among tribal societies. Comparing across different kinds of military violence leads to a theoretical insight: armies vary in their performance because of the kind of organization used to control fear among their troops. Generalizing the point, we can say that all types of violence fit a small number of patterns for circumventing the barrier of tension and fear that rises up whenever people come into antagonistic confrontation.

The military model also fits police violence during arrests and handling prisoners. Police and military confrontations lead to atrocities by the same path: the sequence of emotional events that, in chapter 3, I call “forward panic.” Crowd violence or riot also resembles military violence in some central mechanisms; much of the time confrontation is largely bluster and gesture but leads to little real harm; what is fateful are sudden breaks in the solidarity of one side, which spreads them out into an open field of small groups, where a superior number from one side can isolate and beat up an individual or two separate from their comrades. These are all very ugly forms of violence when we actually see them in detail; indeed, the disparity between their idealized self-image and their atrocious reality is one more situational feature they have in common.

These various forms of violence are subtypes within one of the main pathways around confrontational tension and fear: find a weak victim to attack. Domestic violence is harder for outside observers to study directly, and recordings are virtually nonexistent; we rely here on interview reconstructions, which are limited by being largely confined to reports from just one participant. Nevertheless, working through a large body of evidence, I conclude that the major forms of domestic violence resemble the type of military and police situation that fits under the rubric of “attacking the weak.” The nastiest version of this happens when the confrontational tension builds high, followed by a sudden collapse so that an opponent who initially seems threatening or frustrating turns out to be helpless, unleashing in the other a transformation of fear and tension into ferocious attack. There are also more institutionalized forms of attacking the weak, repetitive patterns in which one or both sides become accustomed to acting out the roles of strong and weak in a situational drama. These include bullying and also the kinds performed by specialists in criminal violence, muggers and holdup artists, who have perfected their skills at finding the right kinds of victims in the right kinds of situations; their success depends upon battening upon confrontational tension itself. Comparisons across disparate forms of violence thus uncover similar mechanisms of emotional interaction.

In another large set of situations, a very different pathway circumvents situational tension and fear. Instead of finding a weak victim, the focus of emotional attention is on the audience before whom the fight is performed. These fights differ greatly from the attacking-the-situationallyweak kind of pathway, because the fighters attend much more to their audience than to each other; as we shall see from evidence presented in chapter 6, the stance of the audience has an overwhelming effect on whether and how much violence is carried out. Such fights are typically stylized and limited, although what happens within those limits may be bloody enough or indeed deadly; in one major variant, violence is socially organized as fair fights, limited to certain kinds of appropriately matched opponents. Here again the social structures promoting and controlling such fights best become visible by comparing across situations. These include personal fights as observed on streets or places of entertainment; fighting as a form of carousing fun; children’s ordinary scuffling and mock-violence; dueling; martial arts and other fighting schools; and sports violence among both players and fans. This set of situations might be regarded as violence for fun and honor, in contrast to the truly nasty forms of violence noted earlier, which depend upon finding a situationally weak victim. Nevertheless, as we look into the micro-realities of fighting for fun and honor, we find they too remain shaped by confrontational tension and fear; people are still for the most part not good at violence, and what they manage to do depends on how attuned they are with an audience that gives them emotional dominance over an opponent.


The most common pathway around confrontational tension and fear is a very short one, leading no further: people do not get beyond the emotional tensions ofthe confrontation, but confine themselves to bluster, or to finding face-saving or sometimes humiliating ways of backing down. When violence does break out, it is usually incompetent, because tension and fear remain during the performance.

One reason that real violence looks so ugly is because we have been exposed to so much mythical violence. That we actually see it unfolding before our eyes in films and on television makes us feel that this is what real violence is like. Contemporary film style of grabbing the viewers’ attention with bloody injuries and brutal aggressiveness may give many people the sense that entertainment violence is, if anything, too realistic. Nothing could be farther from the truth. The conventions of portraying violence almost always miss the most important dynamics of violence: that it starts from confrontational tension and fear, that most of the time it is bluster, and that the circumstances that allow this tension to be overcome lead to violence that is more ugly than entertaining. The entertainment media are not the only sources of pervasive distortions of the reality of fighting; the verbal conventions of bragging and threat, and of telling stories about fights we have seen, all contribute to making violence a modern-day mythology.

A particularly silly myth is that fights are contagious. This is a staple of old film comedies and melodramas. One person punches another in a crowded bar or restaurant; the waiter topples over with a tray, outraging another patron, and in the next frames everyone is hitting everyone around them. This fighting of all against all, I am quite certain, has never occurred as a serious matter in real life. The typical response of bystanders when a fight breaks out in a crowded place is to back away to a safe distance and watch. Polite middle-class crowds react with more unease or horror, shrinking away as far as possible without showing overt panic; I have witnessed this, for instance, when a couple of homeless men got into a scuffle on the sidewalk outside a downtown theater while the audience was outside during intermission. The punch-throwing was brief, followed by the usual hostile muttering and gesturing; the well-dressed middle-class people kept their wary distance in hushed unease. In boisterous working-class or youth scenes, the crowd generally will make a space for the individuals to fight; it sometimes cheers and shouts encouragement from a safe distance. But if the level of anger is high among the principals, onlookers tend to shrink back vocally as well as physically.5 Even more so with fights in sparsely populated public places: bystanders keep their distance.

What one does not see is a contagion of belligerence, everyone starting to fight with everyone else. People are not on a hair trigger of aggressiveness, ready to be released by the slightest catalyst. The Hobbesian image of humans, judging from the most common evidence, is empirically wrong. Fighting, and indeed most overt expressions of conflict, most typically call out fear or at least wariness.

The exception to non-contagiousness is when the crowd is already divided into antagonistic group identities. If a fight breaks out between individuals from opposing groups, others from their side may join in, and the fight will expand. This is one typical scenario by which crowds of rival football supporters (so-called football hooligans, especially British) begin their violence; it is also an instigator of ethnic violence and other kinds of what Tilly (2003) calls “boundary activation” of collective identities. This is not war of all against all; the inaptly named “free-for-all” may look chaotic and unstructured to outsiders, but it is indeed quite strongly organized. This organization is what enables individuals to overcome the pervasive fear that keeps most of them from fighting; if it were not socially well organized, wide-participation fighting would not be possible.

Even in these instances, we should be wary of assuming that all confrontations of individuals belonging to hostile groups lead to mass participation. Football hooligans in a strange town, encountering the local supporters, may shout insults, threaten, even skirmish a little, running forward and then back into the safety of their side; but in many instances they do not get into a full-scale “aggro.” The catalytic moment does not always happen; the participants on both sides are often content to find excuses, especially when they are overmatched, or even just equally divided; the confrontation they are looking for, they decide, is still in the future. Such mini-confrontations play a considerable part in the ongoing lore of the group; it is what they like to talk about, what their conversational rituals center on in their drinking gatherings, as they reinterpret the events of the past hours or days; the standoff is often blown up into a battle, or taken as a sign of the other side’s cowardice in backing down from their toughness (King 2001; Eric Dunning, personal communication March 2001). Groups that engage in some fighting build up mythologies around themselves, exaggerating the amount of fighting and their performance in it, and downplaying their own tendencies to back away from most fights.

Another apparent exception to the non-contagiousness of fighting is the friendly free-for-all such as pillow fights or food fights. Pillow fights at a festive occasion such as a children’s sleep-over typically have an all-against-all character; this both promotes and enhances a mood of hilarity, implying that the situation is very unusual, framed as an exceptionally good joke. The multi-sidedness of the pillow fight is more of a participation-spreader, bringing everyone into the collective fun. In this respect, friendly pillow fights are like New Year’s or other carnival celebrations, which involve throwing streamers and blowing noise-makers indiscriminately at other people. The same goes for bathers playfully splashing water at each other in a swimming pool—by my observations, this occurs in the early moments, as soon as a group of acquaintances has entered the pool, that is, entered the festive space. Nevertheless, if the play turns at all rough, it falls into a two-sided pattern. Pillow fights taking place as a form of entertainment in prison cells, for example, often escalate by putting books or other hard objects inside the pillow cases, and these turn into ganging up on the weakest victim, the one most prone to break down (O’Donnell and Edgar 1998a: 271). During food fights in institutional dining halls, people throw food around more or less randomly, without looking at targets; they throw it up in the air, generally in the direction of persons at distant seats, or better yet, distant tables. In these settings, food fights have both the character of spontaneous self-entertainment, but also a revolt against authority in total institutions. Food fights are also observed in popular lunch-time groups in American high schools, but here it is less a free-for-all and more often a form of boy-girl flirting or playful activity among friends, the same persons who engage in food sharing as a sign of intimacy (Milner 2004: chap. 3). The upshot is that we can be pretty sure, when we see a fight in the all-vs.-all mode, that this is only play-violence, not serious; the emotional tone is not confrontational tension-fear, and everyone can sense when it is or isn’t.

A second myth is that fights are long. In Hollywood films (not to mention Hong Kong kung-fu films and similar action adventure films the world over), fist fights as well as gunfights go on for many minutes. Fighters are resilient, taking many blows and coming back to dish them out; crashing over tables, knocking down shelves of bottles, bouncing off walls, falling over balconies and down stairs and hillsides, in and out of cars and other speeding vehicles. Shooting involves much resolute stalking, running from cover to cover, sometimes daringly outflanking the opponent, but never retreating; on the other side, the evil-doers keep coming back, sneakily and warily if not by sheer pugnacity and ferociousness. In the 1981 film Raiders of the Lost Ark, the hero trades punches with a beefy villain for four minutes; then he immediately jumps on a horse to chase and board a speeding truck in another fight sequence, lasting eight and a half minutes. During the course of these sequences, the hero kills or knocks out fifteen of the enemy, plus another seven civilian bystanders. Dramatic time of course is not real time; but whereas most film and stage dramas compress real time to gloss over the dull and routine moments of ordinary life, they expand fighting time by many times over. The illusion is further bolstered by fights staged as entertainment. Boxing matches typically are planned for a series of three-minute rounds, up to a maximum of thirty or forty-five minutes of fighting (in the nineteenth century sometimes much longer); but these are deliberately controlled by social and physical supports and constraints so as to make most matches produce many minutes of more or less continuous fighting. Even here, referees generally have to prod boxers to stop stalling or tying each other up by clinching. It takes continuous social pressure to keep a fight going. Such fighting is an entirely artificial construction; it is an entertaining spectacle precisely because of its extreme departure from ordinary reality.

In reality, most serious fights on the individual or small-group level are extremely short. If we cut out the preliminaries and the aftermath, with their insults, noise, and gesturing, and look only at the violence, it is often remarkably brief. The actual gunfight at the O.K. Corral in Tombstone, Arizona, in 1881 took less than thirty seconds (see reprint of Tombstone Epitaph, Oct. 1881); the 1957 movie version took seven minutes. Crimes involving the use of guns almost never take the form of gunfights between sides both armed and firing at each other. The vast majority of murders and assaults with deadly weapons consist of one or more armed persons briefly attacking an unarmed person. Since the latter half of the twentieth century, gang fights, drug turf battles, or reputational confrontations, as in violence-prone areas such as inner-city racial ghettos, often involve guns. But they are usually not gun-battles, but very brief episodes, usually with only one side firing.

Fist fights are also generally brief. Many bar-room brawls and street fights are one-punch affairs. The lore of such fighting is that whoever gets in the first punch generally wins. Why should this be so? Consider the alternatives. A two-sided, relatively evenly matched fight could hypothetically go on for some time. But evenly matched scuffling is likely to be unsatisfying when, as is usually the case, neither does much damage, or nothing happens that counts as a dramatic blow producing dominance. Fighters in such situations settle for demonstrating their willingness to fight, and then truncating the actual fighting by letting it degenerate into gesturing and name-calling. Another common occurrence is that one of the fighters hurts himself, such as by breaking his hand in throwing a punch.6 Injuries of this sort are often regarded as fair grounds for ending the fight. A key issue then is when a fight is considered to be over. Far from seeking long-term, knock-down drag-out fights of the Hollywood film or boxing match type, ordinary belligerents are satisfied for fights to be short dramatic episodes, minimizing the period when they are actually fighting. They are willing to give or take a hurt during that period, and then use the hurt as a resolution for the fight, at least for the time being.

A fight of this sort may be part of a series of violent confrontations; for example, a short fight in a bar may lead to one of the participants leaving, getting a gun, and returning to shoot the winner of the first fight. But this is typically two short episodes of micro-confrontation. Individuals’ anger and feelings of being involved in a conflict are not coextensive with their peak capacity actually to carry out violence.

Fights with knives and other cutting weapons also tend to be brief. For the most part, these are situations of flashing knives at each other but letting the confrontation turn into a standoff; where serious injury is done, a quick blow is struck, and the fight is thereby regarded as at an end. Thus another staple of entertainment lore regarding an earlier historical period, the extended swordfight as choreographed in films and plays, was probably for the most part mythical. In early modern Europe, if someone actually succeeded in killing the other, or inflicting serious injury (cases likely to come to the notice of authorities), it was usually described as an ambush, or a group attack on an individual (Spierenburg 1994). This would be the equivalent of the one-punch, sucker-punch bar fight.

There are two important classes of exceptions. Exceptions to a generalization are valuable because they enable us to refine the explanation.

Where individual or small-group fights are prolonged beyond a few moments, it is typically because either (a) the fight is highly circumscribed, so that it is not really “serious,” or it is clearly understood that there are safeguards to limit the fighting; or (b) the type of exception described by the expression “hitting a man when he is down” (although the victim may well be a woman or a child), where in effect there is no real fight but a massacre or punishment.

The typical exception of type (a) has the structure of the boxing match, or even more so, sparring practice for such a match. European aristocrats in the seventeenth and eighteenth centuries spent much time in fencing lessons; nineteenth-century German university students belonged to dueling fraternities, fighting matches that ended not so much in victory as in getting a scar on one’s face as a mark of honor. These are controlled forms of fighting and could be extended for as long as fifteen minutes (Twain 1880/1977: 29–31); not only is the extent of injury generally quite limited, but also the confrontational mood is dampened; these are not angry encounters, but even a form of solidarity.

Just how narrow is this exception becomes apparent when we compare practicing for duels with the duels themselves (for further detail, see chap. 6). Most pistol duels were literally one-shot fights—that is, one shot was planned to be fired by each side. The moment of danger, though real, was brief; if both survived, honor could be considered satisfied. Duels had the same structure as modern fights: typically very brief, within a few seconds of actual violence; preceded by a build-up period of ritual exchange of insults; and terminated by mutual agreement as the result of the conflict, whether by explicit tradition or implicitly.

The same pattern appears in Japan during the Tokugawa era (seventeenth and eighteenth centuries). Samurai were ideally expected to defend their honor with a fight to the death, and could be quite touchy about insults in public places (Ikegami 1995, personal communication). Indeed, samurai went out of their way to make insults very easy, since an accidental knocking of sword scabbards in passing was taken as an affront. One side effect—or perhaps it was the main effect, motivating the practice— was that samurai went about clutching their scabbards, one on each side, the mark and privilege of the samurai rank being to wear two swords. This kept them constantly focused on their emblem of social identity as fighters, even as it prevented most outbursts from occurring. If a fight broke out, it happened on the spot, without the apparatus of challenges, seconds, and advance scheduling found in European duels. Samurai thus tended to be in a constant state of threat and gesture rather than actual fighting. According to the professional lore of the sword instructors, deadly fights should be very brief, consisting of a sudden decisive stroke; in reality, most fighters were probably not this accomplished, although the ideology may have justified the brevity of actual fights. Vastly more time was spent in samurai schools, practicing fighting in controlled ways that obviated injury as well as angry emotion; indeed, such schools tended toward formal exercises of movements aimed at imaginary opponents— like the katas, which make up much of the activity of martial arts schools in karate.

The most famous case of a samurai avenging an insult was the so-called forty-seven ronin, in 1702. One high-ranking samurai was insulted by another over a matter of etiquette in the shogun’s palace; he drew his sword and wounded the insulter, but was quickly disarmed by other attendants. This was not a duel, since the insulter did not draw a weapon; it was not highly effective, since the man was not killed. The incident was apparently very brief, resulting in a few slashes. The attacker was condemned for drawing his sword in the palace, and required to commit seppuku. The forty-seven retainers (ronin) eventually avenged the death of their master, again not by dueling, but by a military assault on the house of the original insulter, killing a number of guards and the samurai lord, who did not defend himself. None of the forty-seven was killed in the attack, indicating that they had overwhelming force, the typical pattern of a strong force ganging up on a weaker one. Even the aftermath did not match the heroic code. The court ruled that avenging honor was not an excuse in this case, but the forty-seven ronin were allowed to commit seppuku as an honorable way to die. Ideally, this was supposed to consist of cutting open one’s bowels with a short knife across the stomach; the agony then would be cut short by being beheaded by a man standing behind the seated samurai. In reality, the forty-seven committed “fan seppuku”—instead of a knife they held a fan with which they gestured a stomach slash, whereupon they were beheaded (Ikegami 1995). It was in fact an execution by beheading, mitigated by the formalities of ritual suicide, which was how the event was publicly announced and received. Japanese samurai movies, continuing an earlier genre of stories, are as mythical as Hollywood westerns.

Another variant on the pattern of prolonged protected fights comes into focus when we examine children’s fights. Fighting among children is the most common form of violence in the family; it is far more common than spousal violence or child abuse (see chapter 4). But children are rarely injured in these fights; in part because children, especially when small, have little capacity to hurt each other in these scuffles. More importantly, children pick their occasions for such scuffles, generally when parents or caretakers are nearby, so that if the fight escalates, they can call for help and end the fight. An example from my ethnographic notes:

Somerville, Mass. December 1994. Family in working-class neighborhood, getting into car on Sunday morning. Father sitting at wheel, warming up car; two boys (about ages 8 and 10) playing around behind car (in alley where it’s parked, outside house), with a little girl (about 3 or 4); mother (woman around 30) coming out of house last. Little girl is getting into back seat from left side of 4-door car; smaller boy bumps her with the door and she starts to cry, whereupon bigger boy hits younger boy, “look what you did!” Mother comes out at just this time; father ignores it. Mother now hurriedly tries to make the boys get into car. They evade her, going behind car and start running around and swiping at each other. Bigger boy has a soft drink sitting on trunk; smaller boy spills it onto ground. Bigger boy now hits him hard and makes him cry. Mother intervenes, threatens the bigger boy, who runs away from her. She turns and puts the smaller boy into the car from the left rear side. The bigger boy now comes and tries to pull him out: “That’s my seat!” Father turns around from front seat and half-heartedly tries to pull one of the boys off. The mother, who starts out hurried but fairly quiet, starts screaming, and pulls the bigger boy out of the car. The bigger boy now appeals to his father, says he’s forgotten something in the house. He goes into house. Now mother demands smaller boy to move over to other side of car; he resists, she finally pulls him out and forces him to move over, protesting that he’s the victim of his older brother. Older boy comes back; same sequence of fighting over back seat, but briefer; finally all get into car (older boy in left back) and car leaves.

In this sense, children act like adults, except that the latter have developed means of bringing fights to an end on their own, whereas children rely on outsiders to do it for them.7 Similarly, fights that break out in schools commonly occur in the presence of a teacher, or where a teacher will likely come quickly to break it up; in prisons, most fights occur in the presence of guards (Edgar and O’Donnell 1998). This is a mechanism by which fights are kept short.

Exception (b) consists in longer violence that may take place in instances where there is overwhelming disparity in force between the two sides, a group gives an isolated enemy a prolonged beating, or a strong individual beats on a weaker. The lesson suggested by this exception is that it is the fighting confrontation rather than violence per se that is hard to sustain for very long: the tension of a one-on-one fight or evenly matched small groups, trading blow for blow, shot for shot; but if one side gets the other down or in an unprotected position, the tension is resolved and violence can proceed.

Real fights are generally short; participants do not appear to have reserves of motivation that carry them into a prolonged violent struggle with another individual. Fights are kept short because participants are good at finding stopping points that they regard as dramatically appropriate. Fights can take longer when they are deliberately staged as not serious, not part of the ground-zero real world. Violent episodes can be longer and more drawn out if they are controlled, restrained both in their likelihood of injury and in their atmosphere of hostility; practice-fighting is thus much lengthier than real fights. Even angry fights tend to happen in places where they can be broken up.

Another entertainment myth is the smiling, joking killer or bad guy. It is extremely rare that killers, robbers, or fighters are in a laughing good humor, or even display sardonic wit.8 The laughing villain image comes across so well precisely because it is unrealistic, giving a coded message that the villainous deeds are not real, but encapsulated within an entertainment frame; hence it is a favored stereotype for cartoons and comic/ fantasy melodrama, and introduces just such a comic tinge into allegedly serious drama. The image enables the viewer to take the entertainment-audience attitude, not the horrified attitude that would occur with real violence. Once again, entertainment violence manages to present violence so that its key feature—confrontational tension and fear—is covered up.


My aim is a general theory of violence as situational process. Violent situations are shaped by an emotional field of tension and fear. Any successful violence must overcome this tension and fear. One way this is done is by turning the emotional tension into emotional energy, usually by one side of the confrontation at the expense of the other. Successful violence battens on confrontational tension/fear as one side appropriates the emotional rhythm as dominator and the other gets caught in it as victim. But only small numbers of persons can do this. This is a structural property of situational fields, not a property of individuals.

As I have argued in a previous book, Interaction Ritual Chains (Collins 2004), emotional energy (which I abbreviate as EE) is a variable outcome of all interactional situations, most of which are not violent. EE varies with the degree that the people present become entrained in each other’s emotions and bodily rhythms, and caught up in a common focus of attention. These are positive experiences when all participants feel solidarity and intersubjectivity. In these successful interaction rituals, individuals come away with feelings of strength, confidence, and enthusiasm for whatever the group was doing: these feelings are what I call emotional energy. Conversely, if the interaction fails to produce entrainment for certain individuals (or if they are subordinated or excluded by others), they lose EE, and come away feeling depressed, lacking in initiative, and alienated from the group’s concerns.

Violent interactions are difficult because they go against the grain of normal interaction rituals. The tendency to become entrained in each other’s rhythms and emotions means that when the interaction is at cross purposes—an antagonistic interaction—people experience a pervasive feeling of tension. This is what I call confrontational tension; at higher levels of intensity, it shades over into fear. For this reason, violence is difficult to carry out, not easy. Those individuals who are good at violence are those who have found a way to circumvent confrontational tension/ fear, by turning the emotional situation to their own advantage and to the disadvantage of their opponent.

It is the features of situations that determine what kinds of violence will or will not happen, and when and how. This means that what happens further back, before people arrive in a situation of confrontation, is not the key factor as to whether they will fight, nor how they will fight if the situation moves in that direction; nor indeed who will win and what kind of damage gets done.


Most existing explanations of violence fall into the category of background explanations: factors outside the situation that lead up to and cause the observed violence. Some background conditions may be necessary or at least strongly predisposing, but they certainly are not sufficient; situational conditions are always necessary, and sometimes they are sufficient, giving violence a much more emergent quality than any other kind of human behavior. As already noted, conditions such as being subjected to poverty, racial discrimination, family disorganization, abuse, and stress are far from determining whether violence will happen or not. This is also true for the venerable psychological hypothesis that frustration leads to aggression, where frustration may be far in the background but also could be quite proximate.

My objection across the board is that such explanations assume violence is easy once the motivation exists. Micro-situational evidence, to the contrary, shows that violence is hard. No matter how motivated someone may be, if the situation does not unfold so that confrontational tension/fear is overcome, violence will not proceed. Conflict, even quite overtly expressed conflict, is not the same as violence, and taking the last step is not at all automatic. This holds as well for a frustration that crops up immediately in the situation: someone may become angry at the frustration and at the person held responsible for it, but that still is not enough to proceed to violence. Many, probably most, frustrated persons swallow their anger, or let it go with bluster and bluff.

It might seem a natural step to form a multi-level theory, combining background and situational conditions. This may eventually turn out to be a good way to proceed. But there is much to be understood before taking that step. Most background theories of violence are concerned with criminal violence in a narrow sense. But there is a lot of violence that is not well understood in terms of background conditions at all: for example, the violence committed by the small proportions of soldiers who are effective fighters, and by rioters, police, athletes and fans, duelists and other elites, carousers and entertainment audiences. Often these violent persons come from the opposite sorts of background as those alleged to be crucial for criminal violence; and these forms of violence have patterns of situational emergence in which emotional dynamics of the group are overwhelmingly apparent. My preferred strategy is to push as far as possible with a situational approach; eventually we may be in a position to work backward and incorporate some background conditions; but I am not yet convinced that is going to be as important as we have usually believed. Here it may be more useful to reverse the gestalt completely, and concentrate on the foreground to the exclusion of all else.

Opportunity and social control theories give a situational emphasis that is surely on the right track. These theories downplay background motives. Generally they assume that motives for violence are widespread; or that motives for transgressions may be situationally emergent. Routine activities theory (Cohen and Felson 1979; Felson 1994; Meier and Miethe 1993; Osgoodet al. 1996), the most prominent version of an opportunity approach, is a theory of crime generally, which is not necessarily violent. In a typical case, the reason a group of youths steal a car may simply be that they found one with the keys left in. Such opportunity explanations leave a much wider gulf to be jumped when the crime is a violent one. The formula for crime is a coincidence in time and space of a motivated offender, an accessible victim, and the absence of social control agents who could deter the crime. The emphasis in routine activities theory is on variations in the latter two conditions, which are held to explain shifts in crime rates apart from any shift in motivating conditions (like the background conditions discussed earlier). What such research has shown is mainly that work and carousing patterns (such as those that involve being out late at night), together with demographic concentrations of particular kinds of persons in particular neighborhoods, affect victimization rates. Since this is an interactive model of several variables, there need not be any change in criminal motivation to account for changes in crime rates; and indeed criminals’ motivations need not be very strong if the opportunities are particularly easy. Although the approach is situational, the analysis mainly focuses on macro-level comparisons. Thus it does not get closely into the process by which violence takes place. The incompleteness of opportunity theory is that it assumes violence is easy; if an opportunity presents itself and no authorities are around to prevent violence, then it pops out automatically. But violence is not easy, and situational patterns of incipient, threatened violence are a barrier that has to be gotten around. The micro-situational mechanism still has to be supplied.

A similar limitation exists with Donald Black’s (1998) theory of the behavior of law. The theory is valid as far as it goes, but where it goes is an explanation of how conflict is managed once it breaks out; varying amounts of formal legal intervention are determined by repetitive, trans-situational features of the social structure: the hierarchical distance among the parties to the dispute, and their degree of intimacy. It is an important theoretical advance to see that moralizing about violence is a variable that can be explained by the participants’ and the social controllers’ locations in social space. But the theory still assumes that violence is easy; its focus is on what happens after violence breaks out, on the societal reaction. It is true, for example, that much violence is self-help, the escalation of ongoing conflicts among persons known to each other, and that the very intimacy of the relationship deters formal intervention by police and legal authorities. But self-help violence still needs to be situationally constructed; it still needs to get past the barrier of confrontational tension and fear. This is not easy; and there is not as much self-help violence as might be expected from the numbers of persons who have motives to help themselves against local antagonists (described, for example, in unpublished research by Robert Emerson, UCLA, on roommate quarrels).

A similar problem exists with more macro explanations of violence, which includes theorizing violence as resistance. Resistance theories frame violence as a local response to subordinate location in large-scale social structure; usually this is class location in the capitalist economy, sometimes abstracted more generally into a structure of domination that includes race and gender.9 The micro thesis again applies: resistance theory assumes that violence is easy, that all it takes is a motive. But resistance violence is just as hard as any other kind of violence. When resistance violence occurs—or at least violence that plausibly can be construed as such, because it occurs in the lower class or a racial ghetto—it is by going along with situational dynamics and constraints. These are the same patterns found elsewhere: small numbers of violent specialists, getting their energy from the unviolent part of the group, requiring the support of audiences, and battening on the emotionally weak. Micro-situational conditions favor attacking victims inside the community of the oppressed, much more than its ostensible class oppressors. Resistance theory often has a twisted quality: as an interpretation put forward by altruistic outsiders bending over backward to be sympathetic, it heroizes and justifies violent predators who perform most of their violence against the members of their own oppressed group.

Even in those instances where violence is most explicitly resistance, as in ghetto uprisings under slogans of rebellion against racial injustice, the violence is almost all local, and most of the destruction is in one’s own neighborhood. The rhetoric of the uprising is one thing, the actual violence is another; the attacks are local because this is the situationally easiest way. When an ideologically aroused group invades someone else’s neighborhood, it is less likely to be vertical resistance to the overarching social order, but rather a lateral assault on some other ethnic group, thus forfeiting the moral legitimacy of being seen as resistance by altruistic onlookers from higher social classes.

Cultural explanations of violence are almost always macro explanations; a wide-ranging, trans-situational culture is assumed to be the (necessary, and implicitly even sufficient) explanation of why the violence takes place. There is the same flaw here, from the point of view of micro-situational analysis, as in resistance theories, even when the explanation is turned around. Some theories regard violence, not as resistance, but as imposition from above, disciplining and deterring resisters in the name of upholding the cultural order. Thus a culture of racism, homophobia, or machismo is offered as an explanation of attacks on minorities, women, and other victims. This kind of interpretation at least is on stronger empirical grounds than resistance interpretations, since such attackers usually vocally state their prejudices during their attack, whereas the alleged resisters usually do not. But the interpretation suffers from failing to look closely at the dynamics of micro-situations; the great majority of them involve bluster and bluff, substituting verbal insult for actual violence, and sometimes (given additional conditions) using the energy of bluster to carry over into actual violence. It is not at all clear that the insulting expressions used in these situations represent long-held beliefs and deep-seated motivations for action. I will discuss this in greater detail in chapter 8, in regard to the ritual insults used by sports fans and soccer hooligans. There is micro-sociological evidence that racism and homophobia are situationally constructed too. The fact that these words are nouns misleads us into reifying what are actually fluctuating and temporally situated processes.

A similar line of argument applies to “culture of violence” explanations in criminology. Here there is more ethnographic grounding, less imposition of a political interpretation upon the data. But the fact that we can observe distinct groups of persons (such as young men in poverty zones) who talk in a favorable way about violence does not mean that this talk carries over automatically into violent behavior. Violence is hard, not easy. Virtually no cultural discourse admits this; neither perpetrators nor pro-violence groups, nor victims, nor altruistic or righteous observers-from-a-distance. Everyone thinks violence is easy to perform, whether one brags about it, fears it, or hopes to eliminate it. But the micro-situational realities of talking about violence fall into ritual patterns of bluster and bluff, and these rituals provide an ideology that covers up the real nature of violence—that it is hard to perform, that most people are not good at it, including those who are doing the bragging and swaggering. There are cultures of violence in the sense of distinctive networks who circulate this kind of violent talk; but we need to get beyond taking them at their word.

Macro-cultural approaches to violence become vacuous when they reach the concept of “symbolic violence.” This helps us not at all to explain real violence, but muddies the analytical task. Physical violence has a clear core referent, which we can study using micro-situational observations. We are in a very different conceptual universe, when Bourdieu writes of schooling requirements as symbolic violence, and in general invokes the entire arena of symbolic possessions as “the gentle, invisible form of violence, which is never recognized as such, and is not so much undergone as chosen, the violence of credit, confidence, obligation, personal loyalty, hospitality, gifts, gratitude, piety. . . . [S]ymbolic violence is the gentle, hidden form which violence takes when overt violence is impossible” (Bourdieu 1972/1977: 192, 195). This is a merely rhetorical usage, a way of dramatizing the argument that school achievement, cultural tastes, and ritual practices are part of a self-reproducing structure of stratification, which the author wants to impress on his audience is morally illegitimate. But the dynamics of school requirements and cultural stratification are not at all similar to the dynamics of physically violent confrontations. The latter is a micro-situational process revolving around emotions of fear, tension, and forward panic, with strong elements of emergence; Bourdieu’s “symbolic violence,” to the contrary, is smooth, tension-free, non-confrontational, highly repetitive, and without situational contingencies.10

Of course, any core concept has its borderline areas. It is not useful to insist that violence must fit an exact preconceived definition. When people aim blows or weapons at each other, there is a period of buildup and anticipation, and these periods are worth studying even when they do not lead to actual violence. As we know, blows and projectiles often miss their targets; sometimes they are not very much intended to hit; sometimes they hit someone unintentionally. Where do we draw the boundaries? Are threats a form of violence? Clearly they are close enough to it that we have to put them into the model of situational dynamics. And this is so even though there is a good deal of cursing which does not lead to violence. By the same token, we will study the situational dynamics of quarreling, and of fearful, tense, and hostile emotions generally. The methodological rule should be to let the research process find its own borders. By this criterion, rhetorical pseudo-explanations get ruled out because they do not connect.

“Symbolic violence” is mere theoretical word play; to take it literally would be to grossly misunderstand the nature of real violence. Symbolic violence is easy; real violence is hard. The former goes with the flow of situational interaction, making use of the normal propensities for interaction rituals. The latter goes against the interactional grain; it is because the threat of real violence runs counter to the basic mechanisms of emotional entrainment and interactional solidarity that violent situations are so difficult. It is precisely this tension that produces confrontational tension and fear, the chief feature of micro-situational interaction on which pivot all the features of violence when it does occur.


Finally, a few words about a prominent research program that has a very explicit theory of violence, evolutionary psychology. This theory extrapolates from a general theory about evolutionary genetics to specific human behaviors including homicide, fighting, and rape (Daly and Wilson 1988; Thornhill and Palmer 2000). The theory makes much of the empirical patterns that young men of peak reproductive ages perform most violence, and that the instigation of violence is often sexual jealousy or masculine posturing. Violence is interpreted as an evolutionary selected propensity for males to struggle over reproductive dominance.

One cannot rule out a priori the possibility that there are genetic components of human behavior. But a wide range of empirical comparisons lead to the conclusion that the genetic component, if it exists, is small, and is overwhelmed by social conditions. For one thing, violence is not confined to young men of reproductive age. The most common type of violence in the family, for example, is not between adult sexual partners; this is outweighed by parent/child violence, typically in the form of severe corporal punishment; and that in turn is less frequent than violence among children (see chapter 4). Violence among children is not very severe, for reasons that we will consider, including the tendency for violence that is restricted and regulated by outsiders (in this case, adults) to be chronic rather than severe. This poses a puzzle for evolutionary theory; children scuffling starts at quite young ages, and often involves aggression by little girls, which is gradually restricted as they get older (Trembley 2004). In sheer quantity, the greatest frequency of incidents of violence occurs at non-reproductive ages, and is not exclusively intra-male. Evolutionary psychologists may overlook this kind of violence because it is not very severe, and doesn’t get recorded in the official crime statistics; nevertheless a comprehensive theory should account for all sorts and all levels of intensity of violence. Micro-situational theory does quite well in incorporating data on children; as we shall see, scuffling among small children shows the same two patterns that are at the center of adult violence: the situationally strong ganging up on the weak and fearful, and staged, limited fights. The pattern is structural rather than individual; taking children out of the group and putting in others rearranges the pattern of dominance, and shifts who plays the roles of bully and victim (Montagner et al. 1988).

Evolutionary psychology is also vulnerable on its main turf, the propensity of young men to be involved in serious violence. It is not difficult to construct alternative explanations of why young men are violent, based on social conditions. Of any age group, they have the most ambiguous status in society; physical strength and violence is the one resource in which they have superiority, whereas they rank low in economic position, deference, and organizational power. The point I want to stress again is my micro-sociological refrain: evolutionary theory assumes that violence is easy—provided that the genes are primed for it—whereas in fact violence is hard, even for young men. Indeed, the majority of our micro-evidence is about the failures of violence among young men.

Large sectors of the intellectual world today dismiss evolutionary theory: partly in response to its perceived insensitivity to cultural and interactional patterns; partly out of long-standing intellectual antagonism between interpretive and positivistic approaches, between Geisteswissenschaft and Naturwissenschaft. Although my intellectual alliances are largely with the interpretive camp, nevertheless I want to cross over to evolutionist terrain and suggest that evolutionary psychology has made two serious mistakes, on its own terms.

The first mistake is about what has genetically evolved. The evolutionary orthodoxy of today holds that humans have evolved to be egotistical gene propagators, and that males have evolved the biological hard-wiring to be aggressive in order to propagate their genes in preference to some other male’s genes. I suggest a very different interpretation of what is the main evolutionary heritage on the biological level. As I have argued elsewhere (Collins 2004: 227–28, in the context of explaining human eroticism), humans have evolved to have particularly high sensitivities to the micro-interactional signals given off by other humans. Humans are hard-wired to get caught in a mutual focus of intersubjective attention, and to resonate emotions from one body to another in common rhythms. This is an evolved biological propensity; humans get situationally caught up in the momentary nuances of each other’s nervous and endocrinological systems in a way that makes them prone to create interaction rituals and thus to keep up face-to-face solidarity. I am making more than the banal point that humans have evolved with large brains and a capacity for learning culture. We have evolved to be hyper-attuned to each other emotionally, and hence to be especially susceptible to the dynamics of interactional situations.

The evolution of human egotism, then, is far from primary; it emerges only in special circumstances, for the most part rather late in human history (see Collins 2004, chap. 9, “Individualism and Inwardness as Social Products”). All this has a direct effect on human violence, although rather the opposite of the premises of evolutionary psychology. Humans are hard-wired for interactional entrainment and solidarity; and this is what makes violence so difficult. Confrontational tension and fear, as I will explain in greater detail, is not merely an individual’s selfish fear of bodily harm; it is a tension that directly contravenes the tendency for entrainment in each other’s emotions when there is a common focus of attention. We have evolved, on the physiological level, in such a way that fighting encounters a deep interactional obstacle, because of the way our neurological hard-wiring makes us act in the immediate presence of other human beings. Confrontational tension/fear is the evolutionary price we pay for civilization.

Humans have the capacity to be angry and to mobilize bodily energies to be forceful and aggressive. These, too, have physiological bases; they are universal in all societies (Ekman and Friesen 1975), and are found among most small children.11 The capacity for anger is explained in evolutionary psychology as a means of mobilizing bodily effort to overcome an obstacle (Frijda 1986: 19). But when the obstacle is another human being, the hard-wired capacity for anger and aggression meets an even stronger form of hard-wiring: the propensity to become caught up in a shared focus of attention and the emotional rhythms of other people. How do we know the propensity for interactional entrainment is stronger than mobilized aggression? Because the micro-situational evidence, reviewed throughout this book, shows the most frequent tendency is to stop short of open violence; and when violence does emerge, it happens in an interactional process that is oriented in detail to overcoming the confrontational tension, while continuing to leave traces of it.

That is not to say that humans cannot be in conflict. They often have conflicting interests, and they often express their antagonism toward opponents. But this antagonism is expressed for the most part against other persons (or better yet, vaguely indicated groups) who are at a distance, preferably out of sight and hearing. It is the immediately situational confrontation that brings up an overwhelming tension; for face-to-face violence to occur, there must be some situational way around this emotional field.

Here I will introduce a second feature of evolution that is relevant to the construction of human violence. Now we are concerned not with biological evolution of the physical hard-wiring of human bodies, but with human institutions, which can also be seen as evolving over time, with some institutions being selected for survival and others selected out. If humans beings have evolved on the physiological level to be full of confrontational tension when they encounter another human in an antagonistic mode, the development of violence in human history must be due to the social evolution of techniques for overcoming confrontational tension/fear.

Historical comparisons show that social organization is a huge component in determining the amount of violence that takes place. The history of armies is the history of organizational techniques for keeping men fighting, or at least not running away, even though they are afraid. In tribal societies, battles are short, mostly skirmishes among a few hundred men or less, intermittently for a few hours, usually ending when a single victim is killed or seriously wounded. Without social organization to keep soldiers together in ranks, they dart back and forth across a skirmish line, a few men at a time, running away if they are in enemy terrain for more than a few seconds. The structure is analogous to today’s gangs who carry on vendettas in the form of reciprocal drive-by attacks, firing at the opposing group from a passing car: when one group meets the other en masse, they generally bluster and insult but contrive to evade an open clash. The comparison shows that evolution of social technique for promoting violence is not just a matter of historical time; groups within modern societies are in the same structural condition as small primitive tribes, without the organizational apparatus for compelling troops to stay in a combat situation.12

More complex social organization in ancient Greece, Rome, and China brought larger numbers (sometimes on the order of tens of thousands) and more disciplined troops into battle and could keep them in combat as long as a day. One day was also the normal length of battles in medieval Europe. By the time of the Napoleonic wars, armies were sometimes on the order of hundreds of thousands of men, and battles lasted as long as three days. In the world wars of the twentieth century, battles were sustained as long as six months or more (e.g., Verdun, Stalingrad), backed up by a massive bureaucratic apparatus. In all historical periods, most of the troops were young men around peak reproductive age, but what determines the amount of killing done is the kind of social organization. Struggle for reproductive fitness does nothing in explaining the variance. What has evolved have been the organizational techniques for keeping soldiers in line where they could do some damage (or at least to stand up to long-distance weapons that would do damage to them). These techniques have evolved through such devices as the close-ranked phalanx; parade-ground drilled troops surrounded by an officer corps concerned with keeping them in line; the politicized appeals and morale-building techniques of modern mass armies; bureaucratic methods for entrapping individuals in an inescapable organization; as well as coercive specialists like military police whose job is to keep soldiers from running away (Keegan [1976] may be read as a comparison of such techniques across several historical periods; see also McNeill [1982, 1995]).

Military organization is the easiest place to trace the social techniques for overcoming our biological propensity not to be violent. There are other spheres of violence where techniques have evolved, such as the evolution of dueling, martial arts and other fighting schools, and of sports fans’ collective behavior routines. The development of football hooligans in twentieth-century Britain, for example, may be seen as the evolution of techniques beginning with participation in the staged excitement of sports contests, then emancipating the excitement away from the game itself so that an elite of specialists can promote their own form of “riot on demand.” These themes will be taken up in later chapters.

The old-fashioned usage of “evolutionary” to mean progress does not fit well with the historical pattern of violence; if there is a historical pattern, it is that the capacity for violence has increased with the level of social organization. Violence is not primordial, and civilization does not tame it; the opposite is much nearer the truth. But there is an aspect of evolutionary theory in a technical sense that is relevant here. Its conclusions are not comforting ones. In Norbert Elias’s terms, the pattern can be as much “decivilizing” as it is a “civilizing process.”13 I am not wedded to an evolutionary conceptual vocabulary; I am more inclined to see historical sequences in terms of a Weberian theory of multi-dimensional changes in the social organization of power (on this, the most comprehensive formulation is Mann [1986, 1993, 2005]). Techniques for carrying out violence must always be fitted to the task of overcoming confrontational tension/fear; however extensive these organizations are at the macro and meso level, their effectiveness is always tested at the micro level. What the evolutionary perspective does for us here is chiefly to remind ourselves of a very long-run perspective; it is the biological hard-wiring of human beings to have so much emotional difficulty at face-toface violence that has set the problem which the development of social techniques has tended to solve. Fortunately for human welfare, the problem to a large degree still resists solution.


The book is theoretically organized but strongly oriented toward the data. It aims to depict violence at as close range as possible. I have pressed into service every source of information that has been accessible to me. I have tried to exploit visual records wherever possible. Video recordings of fights are chiefly accessible for police, sports, and crowd violence. Video is occasionally useful on contemporary warfare; more revealing is anthropological film of tribal war. Still photos have turned out to be even more useful than video tape, since they can catch emotion and show the details of bodies in space. I insert photos in the text as much as possible within practical constraints. Some of my generalizations draw on my entire photo collections of particular kinds of violence.

Another major source is observation. I have made use of my own observations wherever there was something to gleaned from them. Some of these were deliberately gathered, when I was in violent zones at dangerous times (living in certain parts of East Coast cities has facilitated this), or by police ride-alongs; others have come from being alert, ready to drop into a sociological mode and to look carefully and make notes when something comes up. This is not as melodramatic as one might think; I am interested in conflict situations at low levels as well as high, and it is of interest to see how people handle confrontations, most of which do not in fact escalate all the way to violence, let alone extreme violence.14

On some topics in this book, I have made extensive use of student reports. These are retrospective accounts of situations that my students have observed. I have primed them by instructing what to pay attention to: emotions, body postures, the details of timing. I asked them to describe a conflict they have seen up close, which did not have to be violent; the corpus includes quarrels and abortive fights, an important part of the range of situational dynamics. Given that these students are largely from middle-class backgrounds (although widely ranging in ethnicity and country of origin), the kinds of violence they report tend to be limited to carousing, entertainment, and sports settings, with a certain amount of domestic conflict, and some descriptions of demonstrations and riots. Obviously such data cannot be used to count the statistical frequency of various kinds of violence; but they are very revealing on the relationship among different features of situations, which is what I am after.

I have interviewed persons who have observed or been involved in violence in various ways: police officers in several countries, ex-soldiers, youth scene musicians, bouncers, judges, and criminals. Throughout my emphasis has been on what they have observed, less upon how they interpret or explain what they see (although one can hardly exclude that). Interviewing has ranged from highly structured (but open-ended) questioning, to informal discussion; where it has been fruitful, I have engaged in lengthy and repeated discussions. It has been particularly useful to ask for observational detail from other ethnographic researchers, who have told me things that go beyond their published reports—not because they were holding anything back, but because I have pushed for material from a new angle of relevance. I have also gotten some detailed accounts of various kinds of violence from court sources. My years of participation in various martial arts schools has also been a source of information.

News reports figure largely on some topics. These vary a great deal in the situational detail they provide; but since violence, especially in its more elaborate forms, is a relatively rare event, there is often no substitute for news accounts. They are especially useful where they report follow-up information in police cases such as ballistics reports. There are also some long stories (such as on riots) available on the Internet, which give much more detail than the truncated news dispatches. Television news reports are usually more cryptic and more commentary-laden, hence less useful, except where they provide videos. The main exception here is on sports violence. I have used my own observations of televised games for analysis of player and fansviolence. American sports are so record-oriented that one can often go from a cryptic news story of a fight to reconstruct much of the context: for example, how the players and teams were doing in the competition leading up to the fight. I have also been able to check certain features such as how frequently players are hit by pitches in relation to when these lead to fights.

Previously published materials are woven into the analysis throughout the book. Some of these are from other researchers; especially valuable are the ethnographers of violence (Elijah Anderson, Anthony King, Bill Buford, Curtis Jackson-Jacobs, Nikki Jones, and others), and those who study the milieux in which certain kinds of violence occur (David Grazian on entertainment scenes; Murray Milner on high school status systems). I am particularly indebted to researchers like Jack Katz who have pioneered in bringing together all the close-up data from various angles. A number of these colleagues (Katz, Milner, Grazian) use collective ethnographies—observational reports from a number of observers collected either retrospectively, or from observers sent out to cover particular scenes. This approach has not been much discussed in the methodological literature, but it has many advantages and deserves wider consideration.

I draw on published interviews (such as with criminals, in or out of jail), as well as biographical and autobiographical accounts of participants in violence (especially military violence). Historians have been useful where they give micro-observational details from their sources.

Literary sources are also sometimes useful. One needs to proceed with care here, since literary accounts of violence are a major source of the mythology that obscures our understanding. This is especially true of film drama, which with tiny exceptions are extremely unreliable depictions of violence. Some literary accounts, chiefly in the naturalistic style of the early twentieth century, are useful for detail on warfare and on fights, and for the micro-dynamics that lead up to fights, or for the carousing scenes that are their background. A few writers, like Tolstoy, Hemingway, and Fitzgerald, were micro-sociologists before the occupation was invented. Older literature such as Homer and Shakespeare, mythology-propagating in other respects, nevertheless is sometimes useful for describing the ritualism surrounding violence in particular historical periods if not the process of violence itself.

Quantitative data is also used here where relevant. It has been useful (though hard to get) on certain aspects of police violence; and military reconstructions have been at the heart of the academic awakening of how violence really happens, including counts of soldiers firing, hits, ammunition expended, and casualties. A few demonstrations and their casualties (such as the 1970 Kent State national guard killings) have been reconstructed in detail; and I have drawn on data on looting, arrests, and time-patterns of the spread and severity of riots.

Throughout I follow the rule to make my own interpretations of the data. This often means detaching them from the reporter’s or the previous analyst’s concern for what is important, and from their framework of understanding. One might say that sociology is to a large extent the art of reframing other people’s observations. Where the observations are those of previous sociologists and the reframing is strongly overlapping, we can speak of cumulative theoretical progress.

My sources are very heterogeneous. This is as it should be. We need as many angles of vision as possible to bear on the phenomenon. Methodological purity is a big stumbling block to understanding, particularly for something as hard to get at as violence. Obviously, the micro-sociological study of violence can be done better in the future than as I have done here; for now, it is the direction of movement that counts.


Chapter 2 lays out the basic model: violent situations are full of confrontational tension and fear. Hence, most violence is bluster and standoff, with little actually happening, or incompetent performance with mostly ancillary and unintended damage. For real harm to be inflicted on the enemy, there must be pathways around confrontational tension/fear; what these are is mapped out in the following chapters.

Chapter 3 describes a special kind of dynamic sequence when a tense confrontation is suddenly resolved in favor of one side, which takes overwhelming superiority. The result is what I call a forward panic. Many famous atrocities (including many in the headline news) come about in this way.

Chapters 4 and 5 examine the pathways around confrontational tension/fear that consist in attacking a weak victim. Here we look at the situational dynamics of domestic violence, bullying, mugging, and holdups. Some of these are more institutionalized than others, going on repetitively over time. Forward panic, treated in the previous chapter, is also a variant on attacking the weak, although at the other end of the continuum, where the weakness is emergent and the suddenness of the emotional shift is the key to the ferocity of the attack. All these forms of attacking the weak show a key feature of successful violence: picking a target that is emotionally weak, which is more important than being physically weak.

So far, these chapters deal with violence that is ugly and morally despicable, once we see it close up. The second part of the book covers a different set of pathways around confrontational tension/fear. Here the violence is honorable, happy, ebullient, or at least in an in-between zone where it is socially excused and covertly encouraged. Chapter 6 deals with fighting that is deliberately staged for an audience; the same features that make it limited and protected also tend to elevate its practitioners into the sphere of an honored elite. Even here, confrontational tension/fear remains and shapes the violence, like the return of the repressed.

Chapter 7 considers various ways that happy occasions of celebration, carousing, and entertainment can give rise to violence; as well as how some kinds of unhappy violence, such as riots, can take on a carousing tone.

Chapter 8 explains how the structure of sports as dramatic pseudo-violence gives rise to real violence at predictable moments among players and fans. I consider also the conditions under which fans’ violence spills over outside the sports arena, and even becomes autonomous of it: the “B-team” promoting itself to equal or superior status to the “A-team” in the emotional dramatics of sporting action.

Chapter 9 looks at how fights do or do not start. I focus in on the micro-dynamics of bluster and bluff, and examine how these may be institutionalized as a preferred style in the inner-city code of the street.

Chapters 10 and 11 consider who wins and loses fights as a process of micro-situational domination. Success in violence is stratification of an emotional field, parallel to the “law of small numbers” that shapes creativity in intellectual and artistic fields; all these are variants on seizing emotional dominance over limited niches in an attention space. Those who become the violent elite—“elite,” of course, in a structural sense, which may be morally despised as well as adulated—get their emotional dominance from all the other persons in the field. They batten emotionally on their victims, drawing their success from the same process that makes their opponents fail; and they capture the emotional energies of lesser members of the supporting cast and audience.

There is at least this much of a sociological silver lining. Violence has very strong structural limits, by its very nature as the product of an emotional field. The same features that make a minority of persons successfully violent makes the rest of us unviolent. What we can make of this pattern constructively for the future remains to be seen.


Since we social scientists tend to be polemical, and to act as if our own theoretical approach is the only correct one, I would like to go on record as saying that micro-sociological theory is not the whole of sociology. Researchers have successfully studied large-scale structures—networks, markets, organizations, and states and their interplay in the world arena—without looking at micro-details. We have cumulated some useful theories about these meso and macro structures, and I am not suggesting that sociologists should throw these aside to concentrate only on face-toface situations. The issue isn’t ontological—what is real and what isn’t— but pragmatic: what works and what doesn’t. In the particular area of violence research, perhaps more than any other topic, we have misunderstood the most basic micro-interactional pattern. We have assumed that violence is easy for individuals to carry out, so we skip the micro level as unproblematic and turn to conditions in the meso background or the macro organization or overarching culture.

This turns out to be a pragmatic mistake. Violence is not easy, and the key stumbling blocks and turning points are at the micro level. That does not mean that meso and macro conditions don’t exist, or that they cannot be usefully integrated into a more comprehensive theory, once we get the micro mechanisms right.

This book may strike many readers as altogether too micro. It cuts out preceding motivations, background conditions, and long-term consequences of violence. It also omits the way in which violence is produced by larger social structures than the immediate situation, such as by militaries or politics. I agree. But in order to focus closely on the micro-dynamics of violence, it is necessary here to bracket the rest. This book is the first of a two-volume series. The second expands the frame to what has been left out. Among other things, it considers what we know about institutionalized violence, or rather, that which is repetitive, structured, and thus organized into meso-and macro-organizations that provide a regular flow of resources for specialists in violence. It will consider such topics as war and geopolitics, as well as torture and the many contexts and varieties of rape.

This expansion of the topic of violence pushes across several conceptual and empirical boundaries. The topic of large-scale and long-term structures for producing violence borders on the theory of conflict generally; this is a larger topic, since conflict is often not violent. The two are connected by a process of escalation and counter-escalation, which I will broaden to include the crucial but less often considered theory of de-escalation. The second volume will focus on conflict—violent and unviolent alike—as a process swelling up and ebbing away in time. It will attempt to map out the time-laws of when and how conflict occurs at some moments and not in others. This will make time-process a key feature of violence in its own right, apart from other conditions that promote violence; the occurrence of violent events depends on its timing in relation to other such events, as well as in the internal flow of timing in micro-incidents. This may take us some further distance toward understanding violence as a relatively rare event, underdetermined by background conditions.

The appropriate relationship of micro-and macro-sociology is not to reduce one to the other, but to coordinate the two levels of analysis where it leads to some useful result. Violence is one area where doing so is crucial. Despite the shift in scale, there is a thread connecting both volumes. This is the theory of the interactional processes of emotional fields, laid out for micro slices of time and space in the present book, and for larger slices in the following one.

In what follows, I will use the male pronouns “he,” “his,” and “him” deliberately to refer to males. There are similarities between males and females in their behavior in violent situations, but the now-conventional expression “he or she” would be highly misleading for this topic. I will discuss female-on-female violence and male/female violence separately and explicitly.

Return to Book Description

File created: 10/31/2007

Questions and comments to:
Princeton University Press

New Book E-mails
New In Print
PUP Blog
Princeton APPS
Sample Chapters
Princeton Legacy Library
Exam/Desk Copy
Recent Awards
Princeton Shorts
Freshman Reading
PUP Europe
About Us
Contact Us
PUP Home

Bookmark and Share