In his 1974 book, Obedience to Authority, Stanley Milgram reports experiments on destructive obedience. In these experiments the subjects are faced with a dramatic choice, one apparently involving extreme pain and perhaps injury to someone else. When the subject arrives at the laboratory, the experimenter tells him (or her) and another subject--a pleasant, avuncular, middle-aged gentlemen (actually an actor)--that the study concerns the effects of punishment on learning. Through a rigged drawing, the lucky subject wins the role of teacher and the experimenter's confederate becomes the "learner."

In the next stage of the experiment, the teacher and learner are taken to an adjacent room; the learner is strapped into a chair and electrodes are attached to his arm. It appears impossible for the learner to escape. While strapped in the chair, the learner diffidently mentions that he has a heart condition. The experimenter replies that while the shocks may be painful, they cause no permanent tissue damage. The teacher is instructed to read to the learner a list of word pairs, to test him on the list, and to administer punishment--an electric shock--whenever the learner errs. The teacher is given a sample shock of 45 volts (the only real shock administered in the course of the experiment). The experimenter instructs the teacher to increase the level of shock one step on the shock generator for each mistake. The generator has thirty switches labeled from 15 to 450 volts. Beneath these voltage readings are labels ranging from "SLIGHT SHOCK" to "DANGER: SEVERE SHOCK," and finally "XX."

The experiment starts routinely. At the fifth shock level, however, the confederate grunts in annoyance, and by the time the eighth shock level is reached, he shouts that the shocks are becoming painful. Upon reaching the tenth level (150 volts), he cries out, 'Experimenter get me out of here! I won't be in the experiment any more! I refuse to go on!" This response makes plain the intensity of the pain and underscores the learner's right to be released. At the 27O volt level, the learner's response becomes an agonized scream, and at 300 volts the learner refuses to answer further. When the voltage is increased from 300 volts to 330 volts, the confederate shrieks in pain at each shock and gives no answer. From 330 volts on, the learner is heard from no more, and the teacher has no way of knowing whether the learner is still conscious or, for that matter, alive (the teacher also knows that the experimenter cannot tell the condition of the victim since the experimenter is in the same room as the teacher).

Typically the teacher attempts to break off the experiment many times during the session. When he tries to do so, the experimenter instructs him to continue. If he refuses, the experimenter insists, finally telling him, "You must continue. You have no other choice." If the subject still refuses, the experimenter ends the experiment.

We would expect that at most only a small minority of the subjects, a cross section of New Haven residents, would continue to shock beyond the point where the victim screams in pain and demands to be released. We certainly would expect that very, very few people would continue to the point of administering shocks of 450 volts. Indeed, Milgram asked a sample of psychiatrists and a sample of adults with various occupations to predict whether they would obey the orders of the experimenter. All of the people asked claimed that they would disobey at some point. Aware that people would be unwilling to admit that they themselves would obey such an unreasonable and unconscionable order, Milgram asked another sample of middle-class adults to predict how far other people would go in such a procedure. The average prediction was that perhaps one person in a thousand would continue to the end. The prediction was wrong. In fact, 65 percent (26/40) of the subjects obeyed to the end.

It is clear to people who are not in the experiment what they should do. The question is, What features of the experimental situation make this clear issue opaque to subjects? Our aim is to suggest some reasons for such a failure of thinking and action and to suggest ways that people might be trained to avoid such failures--not only in the experiment, of course, but in our practical, moral lives as well. What are some of the sources of the failure?

The experimental conditions involve entrapment, and gradual entrapment affects critical thought . One important feature inducing obedience is the gradual escalation of the shock. Although subjects in the end administered 450-volt shocks, which is clearly beyond the limits of common morality and, indeed, common sense, they began by administering 15-volt shocks, which is neither. Not only did they begin with an innocuous shock, but it increased in innocuous steps of 15 volts. This gradualness clouds clear thinking: we are prepared by our moral training to expect moral problems to present themselves categorically, with good and evil clearly distinguished. But here they were not. By administering the first shock, subjects did two things at once--one salient the other implicit. They administered a trivial shock, a morally untroublesome act, and they in that same act committed themselves to a policy and procedure which ended in clear evil.

Surely in everyday life, becoming entrapped by gradual increases in committment is among the most common ways for us to find ourselves engaging in immoral acts, not to mention simply folly. The corrective cannot be, of course, refusing to begin on any path which might lead to immorality, but rather to foresee where paths are likely to lead, and arrange for ourselves points beyondwhich we will not go. One suspects that had the subjects committed themselves publicly to some shock level they would not exceed, they would not have found themselves pushing the 45O-volt lever. We cannot expect to lead, or expect our young to lead, lives without walking on slopes; our only hope is to reduce their slipperiness. Distance makes obedience easier. Another force sustaining obedience was the distance between the victim and the subject. Indeed, in one condition of the experiment, subjects were moved physically closer to the victim; in one condition they had to hold his hand on the shock plate (through Mylar insulation to protect the teachers from shock). Here twelve out of forty subjects continued to the end, roughly half the number that did so when the subjects were farther from their victim.

Being closer to the victim did not have its effect by making subjects think more critically or by giving them more information. Rather it intensified their discomfort at the victim's pain. Still, being face to face with someone they were hurting probably caused them at least to focus on their victim, which might well be a first step in their taking seriously the pain they were causing him.

Both the experimenter's presence and the objective requirements of the situation influenced decisions to obey authority. The experimenter's presence is crucial to the subjects' obedience. In one version of the experiment he issued his commands at a distance over the phone, and obedience was significantly reduced--to nine out of forty cases. The experimenter, then, exerts powerful social influence over the subjects. One way to think about the experimenter's influence is to suppose that subjects uncritically cede control of their behavior to him. But this is too simple. We suggest that if the experimenter were to have told the subjects, for example, to shine his shoes, every subject would have refused. They would have refused because shining shoes is not a sensible command within the experimental context. Thus, the experimenter's ability to confuse and control subjects follows from his issuing commands which make sense given the ostensible purpose of the experiment; he was a guide, for them, to the experiment's objective requirements.

This interpretation of the experimenter'srole is reinforced by details of his behavior. For example, his language and demeanor were cold--bureaucratic rather than emotional or personal. The subjects were led to see his commands to them as his dispassionate interpretations of something beyond them all: the requirements of the experiment.

Embarrassment plays a key role in decisions to obey authority. The experimenter entrapped subjects in another way. Subjects could not get out of the experiment without having to explain and justify their abandoning their duty to the experiment and to him. And how were they to do this?

Some subjects attempted to justify their leaving by claiming that they could not bear to go on, but such appeals to "personal reasons" were rebutted by the experimenter's reminding them of their duty to stay. If the subjects could not escape the experiment by such claims, then how could they escape! They could fully escape his power only by confronting him on moral grounds. It is worth noting that this is something that virually none of the hundreds of subjects who took part in one condition or another fully did. Failing to address the experimenter in moral terms, even disobedient subjects just passively resisted; they stayed in their seats refusing to continue until the experimenter declared the experiment over. They did not o things we might expect them to--leave, tell the experimenter off, release the victim from his seat, and so on. Why did even the disobedient subjects not confront the experimenter?

One reason seems too trivial to mention: confronting the experimenter would be embarrassing. This trivial fact may have much to do with the subjects' obedience. To confront the experimenter directly, on moral grounds, would be to disrupt in a profound way implicit expectations that grounded this particular, and indeed most, social interaction: namely, that the subject and experimenter would behave as competent moral actors. Questioning these expectations is on some accounts, at least, the source of embarrassment.

Subjects in Milgram's experiment probably did not realize that it was in part embarrassment that was keeping them in line. Had they realized that--had they realized that they were torturing someone to spare themselves embarrassment--they might well have chosen to withstand the embarrassment to secure the victim's release. But rather we suspect that subjects experience their anticipation of embarrassment as a nameless force, a distressing emotion they were not able to articulate. Thus the subjects found themselves unable to confront the experimenter on moral grounds and unable to comprehend why they could not confront the experimenter.

Emotional states affect critical thought. Obviously the emotions the subjects experienced because of the embarrassment they were avoiding and the discomfort produced by hearing the cries of the victim affected their ability to reason critically. We do not know much about the effects of emotion on cognition, but it is plausible that it has at least one effect - a focusing of attention. Subjects seem to suffer from what Milgram has called "Tunnel Vision": they restricted their focus to the technical requirements of the experimental task, for these, at least, were clear. This restriction of attention is both a consequence of being in an emotional state more generally, and it is a strategy subjects used to avoid unwanted emotional intrusions. This response to emotion is, no doubt, a formidable obstacle to critical thought. To reject the experimenter's commands, subjects had to view their situation in a perspective different from the technical one the experimenter offered them. But their immediate emotional state made it particularly difficult for them to do just that: to look at their own situation from a broader, moral perspective.

How can we train individuals to avoid destructive obedience? Our analysis leads to the view that obedience in the Milgram experiment is not primarily a result of a failure of knowledge, or, at least knowledge ofthe crucial issue of what is right or wrong to do in this circumstance. People do not need to be told that torturing an innocent person is something they should not do--even in the context of the experiment. Indeed, when the experimenter turns his back, most subjects are able to apply their moral principles and disobey. The subjects' problem instead is not knowing how to break off, how to make the moral response without social stickiness. If the subjects' defect is not primarily one of thinking correctly, then how is education, even education in critical thinking, to repair the defect? We have three suggestions.

First, we must teach people how to confront authority. We should note as a corollary to this effort that teaching has a wide compass: we teach people how to ride bikes, how to play the piano, how to make a sauce. Some teaching of how to do things we call education: we teach students how to do long division, how to parse sentences, how to solve physics problems. We inculcate these skills in students not by, or not only by, giving them facts or even strategies to remember, but also by giving them certain sorts of experiences, by correcting them when they err, and so on. An analogy would be useful here. Subjects in the Milgram experiment suffered not so much from a failure to remember that as center fielders they should catch fly balls as they did from an inability to do so, playing under lights at night, with a great deal of wind, and when there is an ambiguity about whether timeout has been called. To improve the players' ability to shag fly balls, in game conditions, we recommend practice rather than lectures and the closer the circumstances of practice to the conditions of the actual game, the more effective the practice is likely to be.

Good teachers from Socrates on have known that the intellect must be trained; one kind of training is in criticizing authority. We teachers are authorities and hence can provide practice. Of course, we can only do that if we remain authorities. Practice at criticizing us if we do not respect our own authority is of little use. We do not have a recipe for being an authority who at the same time encourages criticism, but we do know that is what is important. And sometimes we can tell when we are either not encouraging criticism or when we have ceased heing an authority. Both are equally damaging.

Practice with the Milgram situation might help too; it might help for students to"role play" the subjects; if nothing else, doing this might bring home in a forcible way the embarrassment that subjects faced in confronting authority. It might help them develop ways of dealing with this embarrassment. Certainly, it would at least teach them that doing the morally right thing does not always "feel" right, comfortable, natural. There is no evidence about whether such experiences generalize, but perhaps they do.

If they are to confront authority assertively individuals must also be taught to use social pressure in the service of personal values. Much of current psychology and education sees thought, even critical thought, as something that goes on within individuals. But we know better than this. Whether it be in science, law, or the humanities, scholarship is and must be a public, social process. To train subjects to think critically is to train them to expose their thinking to others, to open themselves to criticism from their peers as well as from authority. We insist on this in scholarship because we know that individual thinking, even the best of it, is prey to distortions of all kinds, from mere ignorance to "bad faith."

Further, the support of others is important in another way. We know that subjects who saw what they took to be two other naive subjects disobey, and thus implicitly criticize the action of continuing, were very likely to do so themselves. A subject's sense that the experimenter had the correct reading was undermined by the counter reading offered by the "other subjects." Public reinforcement of our beliefs can liberate us from illegitimate pressure. The reason for this is twofold. Agreement with others clarifies the cognitive issue and helps us see the morally or empirically right answer to questions. But it also can have another effect--a nonrational one.

We have claimed that part of the pressure subjects faced in disobeying was produced by having to deal with the embarrassment that might emerge from confrontation. Social support provides a counterpressure. Had the subjects committed themselves publicly to disobedience before entering the experiment then they could have countered pressures produced by disobedience (during the experiment) by considering the embarrassment of admitting to others (after the experiment) that they had obeyed. Various self-help groups like Alcoholics Anonymous and Weight Watchers teach individuals to manage social pressures to serve good ends.

Social pressures are forces in our lives whether we concede them or not. The rational person, the person who would keep his action in accord with his values, must learn to face or avoid those pressures when they act to degrade his action, but equally important he ought to learn to employ the pressure of public commitment, the pressure implicit in making clear to others what he values, in the service of his values.

Students should know about the social pressures that operate on them. They should also learn how to use those pressures to support their own values. One reason we teach people to think critically is so that they may take charge of their own creations. We do not withhold from engineers who would create buildings knowledge about gravity or vectors or stresses. Rather we teach them to enlist this knowledge in their support.

A second area requires our attention. We need to eliminate intellectual illusions fostering nonintellectual obedience. These are illusions about human nature which the Milgram experiment renders transparent. None of these illusions is newly discovered; others have noticed them before. But the Milgram experiment casts them in sharp relief.

The most pernicious of these illusions is the belief, perhaps implicit, that only evil people do evil things and that evil announces itself. This belief, in different guises, bewildered the subjects in several ways.

First, the experimenter looks and acts like the most reasonable and rational of people: a person of authority in an important institution. All of this is, of course, irrelevant to the question of whether his commands are evil, but it does not seem so to subjects. The experimenter had no personally corrupt motive in ordering subjects to continue, for he wanted nothing more of them than to fulfill the requirements of the experiment. So the experimenter was not seen as an evil man, as a man with corrupt desires. He was a man, like Karl Adolf Eichmann, who ordered them to do evil because he saw that evil as something required of him (and of them) by the requirements of the situation they faced together. Because we expect our morality plays to have temptation and illicit desire arrayed against conscience, our ability to criticize morally is subverted when we find evil instructions issued by someone moved by, of all things, duty. For a fuller discussion of this point, see Hannah Arendt's Eichmann in Jerusalem (1965), where the issue is placed in the context of the Holocaust.

And just as the experimenter escaped the subjects' moral criticism because he was innocent of evil desire, the subjects escaped their own moral criticism because they too were free of evil intent: they did not want to hurt the victim; they really did not. Further. some subjects, at least, took action to relieve the victim's plight--many protested the experimenter's commands, many tried to give the victim hints about the right answers--thus further dramatizing their purity of heart. And because they acted out of duty rather than desire, the force of their conscience against their own actions was reduced. But, of course, none of this matters in the face of the evil done.

The "good-heartedness" of people, their general moral quality, is something very important to us, something to which we, perhaps rightly, typically pay attention. But if we are to think critically about the morality of our own and others' acts, we must see through this general fact about people to assess the real moral quality of the acts they do or are considering doing. A second illusion from which the subjects suffered was a confusion about the notion of responsibility. Some subjects asked the experimenter who was responsible for the victim's plight. And the experimenter replied that he was. We, and people asked to predict what they would do in the experiment, see that this is nonsense. We see that the experimenter cannot discharge the subjects' responsibility--no more than the leader of a bank-robbing gang can tell his cohorts, "Don't worry. If we're caught, I'll take full responsibility." We are all consipirators when we participate in planning and executing crimes.

Those in charge have the right to assign technical responsibility to others, responsibility for executing parts of a plan, but moral responsibility cannot be given, taken away, or transferred. Still, these words--mere words--on the part of the experimenter eased subjects' "sense of responsibility." So long as the institutions of which we are a part are moral, the need to distinguish technical from moral responsibility need not arise. When those institutions involve wanton torture, we are obliged to think critically about this distinction.

There is a third illusion illustrated in the Milgram experiment. When subjects threatened to disobey, the experimenter kept them in line with prods, the last of which was "You have no choice: you must go on." Some subjects fell for this, believed they had no choice. But this is also nonsense. There may be cases in life when we feel that we have no choice, but we know we always do. Often feeling we have no choice is really a matter of believing that the cost of moral action is greater than we are willing to bear--in the extreme we may not be willing to offer our lives, and sometimes properly so. Sometimes we use what others have done to support the claim that we have no choice; indeed, some students interpret the levels of obedience in the Milgram experiment as proof that the subjects had no choice. But we all know they did. Even in extreme situations, we have a choice, whether we choose to exercise it or not. The belief that our role, our desires, our past, or the actions of others preclude our acting morally is a convenient but illusory way of distancing ourselves from the evil that surrounds us. It is an illusion from which we should choose to disabuse our student.