SOCIAL PSYCHOLOGY

The untimely death of Stanley Milgram on December 20, 1984, at the age of 51, marked the end of one of the most productive, eclectic, and innovative careers in social psychology. Although he was best known both among psychologists and the public at large for his landmark studies on obedience to authority (Milgram, 1963, 1974, 1984d), Milgram's span of interest took in a variety of other, less controversial, phenomena. Thus, he left his mark in the form of innovative research on topics as wide-ranging as urban psychology (Milgram, 1970a, 1970b),' the effects of televised antisocial behavior (Milgram & Shotland, 19731, subway norms (Milgram & Sabini, 1978), the lost-letter technique (Milgram, 1969c; Milgram, Mann, & Harter, 1965), the small-world method (Korte & Milgram, 1970; Milgram, 1967b; Travers & Milgram, 1969), crosscultural differences in conformity (Milgram, 1960, 1961), crowds and collective behavior (Milgram, Bickman, & Berkowitz, 1969; Milgram, Liberty, Toledo, & Wackenhut, 1986), mental maps (Milgram, Greenwald, Kessler, McKenna, & Waters, 1972; Milgram & Jodelet, 1976, 1977), the psychology of photography (Milgram, 1976b), the familiar stranger (Milgram, 1977b), and person perception (Milgram, 1984b; see also Colligan, 1980; Herbert, 1985). It is possible to distill from Milgram's work a set of underlying unifying features, themes, and metatheoretical assumptions that, in their totality, add up to a distinctive approach to social psychology. My goal in this article is to present

It is possible to distill from Milgram's work a set of underlying unifying features, themes, and metatheoretical assumptions that, in their totality, add up to a distinctive approach to social psychology. My goal in this article is to present the underlying characteristics of Milgram's approach to the study of social behavior, and then to explore, in detail, the impact of his major research programs, thereby facilitating a fuller understanding of his contributions to social psychology as well as to some other disciplines.

A. A PHENOMENOLOGICAL APPROACH
An understanding of Milgram's work needs to begin with the recognition that Milgram took a phenomenological approach to the study of social behavior. That is, he treated subjective experience as valid data on which to build more systematic analyses of the phenomenon in question. It was this willingness to treat personal experience of, and intuitions about, everyday events seriously that led him into behavioral territories unexplored by others. Thus, for example, in speculating about what might happen if New York subway riders were asked to give up their seats, Milgram and Sabini (1978, p. 33) state "Common sense suggests that it is impossible to obtain a seat on the subway simply by asking for it." Even the obedience studies were ultimately gmunded in everyday observation. Milgram (1963) introduced his first experimental report of his obedience research by linking it to the Holocaust and to earlier writings on authority, obedience, social power, and suggestion. But he concludes his introduction by stating that his work "derives, in the first instance, from direct observation of a social fact; the individual who is commanded by a legitimate authority ordinarily obeys. Obedience comes easily and often. It is a ubiquitous and indispensable feature of social life" (p. 372). We can also see the central role that observation played in Milgram's approach in his article, "The Experience of Living in Cities" (Milgram, 1970a). In it he asks what contribution psychology could make toward understanding city life and how one should proceed empirically. And he answers that "observation is the indispensable starting point. Any observer in the streets of midtown Manhattan will see (i) large numbers of people, (ii) a high population density, and (iii) heterogeneity of population. These three factors need to be at the root of any sociopsychological theory of city life, for they condition all aspects of our experience in the metropolis" (Milgram, 1970a(Milgram, , p. 1461. By allotting a primary role to everyday observation in generating research, Milgram pursued a largely atheoretical approach in most of his studies. This approach was already evident in his doctoral dissertation, in which he criticized social psychology for considering hypothesis testing to be the primary function of experimentation (Milgram, 1960, p. 194). As Tavris (1974b) later put it, "Most psychologists test hypotheses; Milgram asks questions" ( p. 75). When he did invoke theory, he did so usually after the research was conducted. Thus, rather than the more typical hypothesis-generating function that theory plays in social psychology, for Milgram theory came into play in an attempt either to integrate or to explain findings he already obtained. An example of the former is his introduction of the theoretical concept of "overload" in relation to his studies of city versus small-town behavior (Milgram, 1970a) in order to bring "a grab bag of experimental studies . . . under a more rigorous theoretical discipline" (Milgram, 1981b, p. 14). A prime example of the latter, of the introduction of a theoretical concept to serve in an explanatory capacity, is the "agentic state" construct that Milgram introduced after the completion of the obedience experiments in an attempt to explain the behavior of his obedient subjects. According to Milgram, obedience is made possible through the subject's entry into a different experiential state-a state of agency-in which a person no longer views himself as acting autonomously but rather as an agent of an authority. A main feature of the agentic state is a shift in responsibility-the subject sheds responsibility for his actions and hands it over to the person in charge. Actions, no matter how destructive they may be, are no longer guided by the usual moral considerations but rather by how well they carry out the authority's commands (see Milgram, 1974, chaps. 10-12). Milgram first used the agentic-state explanation 7 years after the completion of his obedience studies. It made a brief first appearance in a magazine article by Meyer (1970), in which Milgram is quoted as saying "Just as water can turn to ice under certain conditions of temperature, a person can move to the state of mind that I call agency . . . the critical thing is that you see yourself as the instrument of the execution of another person's wishes" ( p. 132). Even the idea of relinquishing responsibility to an authority as an important concomitant of obedience did not take on a central role in Milgram's thinking until 1967(Milgram, 1967a, although both it and the notion of becoming an agent were already anticipated, though not developed, earlier (see Milgram, 1964b, p. 851;1965c, pp. 72-73). There was, however, no mention of responsibility or of agency in Milgram's (1963) first report on the obedience research.
There is also another sense in which Milgram's approach can be characterized as phenomenological. In studying some facet of social behavior, Milgram would usually present a phenomenological analysis of it that would yield a description of it in terms of its essential, defining properties. Sometimes, in fact, his contribution to our understanding of a phenomenon was purely analytic without involving any empirical research whatsoever. A case in point is an article Milgram wrote exploring the significance of the news (Milgram, 1977f). In it he defines news as "information about events that are going on outside immediate experience" (p. 167). It is a means for extending our eyes and ears to encompass, and prepare us to deal with, a broader range of Occurrences. News can be seen as the consciousness of a society, and totalitarian governments repress the news for the same reason troubled personalities keep certain thoughts out of awarenessbecause of deep-seated insecurities. Milgram notes two further characteristics of the news. First, today news serves more as a source of entertainment than as a source of usable information, and it enables us to indulge our taste for thrills and excitement under the cloak of respectability and enlightenment. Second, news has become increasingly ritualized: the time-slot allotted for the evening news broadcast must be filled even if there is nothing substantive to be communicated.
In a similar vein, Milgram (19778) advances our understanding of another phenomenon-fanaticism-through a conceptual analysis of its defining properties, rather than by empirical means. "Fanaticism," he points out, "is applied to the state of mind of those who are wholeheartedly committed to a set of beliefs and are condemned for it" (p. 58). The fanatic is condemned because his enthusiasm is seen as excessive, a mark of intellectual rigidity and lack of critical judgment, and as a barrier to reasonable persuasion. The maintenance of the social order is predicated on the possibility of resolving conflicts through mutual accommodation and influence. The fanatic is seen as a threat by virtue of his intensity of beliefs, which makes him impervious to influence by the larger society. Milgram notes that fanaticism differs from other forms of deviancy such as criminality. Whereas both society and the criminal recognize that the latter is breaking the law, the fanatic sees the problem in the surrounding society, not in his own behavior and beliefs. From his perspective it is they who "have not grasped the truth and thus are behaving strangely, and even destructively" (p. 59). There is a natural history to every episode of fanaticism, beginning frequently with a drastic change in belief and behavior when the person first "sees the light." This often evokes criticism and attempts to convince the fanatic to return to his previous lifestyle. As a result the fanatic may harden in his belief, his fanaticism may peter out for lack of social support, or a group of like-minded believers may form a community for mutual reinforcement and to attain legitimacy. The larger society might respond to a fanatical group in a number of different ways: It could eventually capitulate to its beliefs, or the group may be isolated in a ghetto, made to emigrate, or be destroyed.
Conceptual analysis also figured importantly as an accompaniment to Milgram's research. Thus, for example, although undoubtedly the quantitative outcome-an obedience rate of 65%-was the most compelling feature of Milgram's (1963) first report on his obedience research, an important adjunct was his analysis of the features of the experiment likely to be promotive of obedience, because it helped explain what at first seemed incomprehensible. Among others, he mentions the reputation of the sponsoring "background authority," Yale University; the fact that the experiment has a laudable purpose, the study of learning and memory; and the sense of obligation on the part of the subject to the experimenter, strengthened by the payment involved, to help the latter complete the experiment.
B. AN EXPERIMENTAL STRATEGY: OUTCOME

MEASURES THAT ARE OBSERVABLE, DISCRETE, AND DICHOTOMOUS
Despite the usefulness of his conceptual analyses, Milgram was, first and foremost, an experimentalist who was able to explore a wide variety of social psychological phenomena ranging from conformity (Milgram, 1960, 196 1, I964a) and obedience  to the lost-letter technique (Milgram, 1969c) and the drawing power of crowds  within the structural confines of the randomized experiment. For Milgram the use of the experimental method was clearly an essential feature of social psychology, as the following quote reveals: "The creative claim of social psychology lies in its capacity to reconstruct varied types of social experience in an experimental format, to clarify and make visible the operation of obscure social forces so that they may be explored in terms of the language of cause and effect" (Milgram, 1977e, p. 1). At the same time, for Milgram the choice of method was dictated by the requirements of the behavior being studied, not vice versa, and when the problem called for it he readily used less powerful research methods. When they were informative enough, he even used simple demonstrations; for example, when wanting to determine how strangers would respond to asking to have their picture taken, his students simply went up to people and asked them "May I take your picture?" (Milgram, 1976b, p. 10).
If there is a unifying characteristic to Milgram's data-gathering techniques, it is not in the types of research designs or techniques he used, but rather in the nature of the target behaviors he studied. In most of his studies, the outcome measure was discrete and dichotomous: It was essentially a yes or no answer to the question of whether a given behavior occurred. Did New York City subway riders give up their seats when requested to (Milgram & Sabini, 1978)? Were lost letters differing in the admirability of their targets mailed or not ? Would strangers, approached either in midtown Manhattan or in a small town, reciprocate an extended hand with a handshake (Milgram, 1977e, pp. 17-lS)? Would viewers of a prime-time TV program, with or without an antisocial act embedded in it, break into and steal money from a charity display (Milgram & Shotland, 1973)? Would a subject commanded to inflict increasingly harsher punishment on a learner obey fully or disobey and stop at some point short of the highest possible voltage (Milgram, 1974)? Even Milgram's other measure of obedience-the specific break-off point on the shock continuum-can be seen as the accumulation of the outcomes of a series of discrete, dichotomous response choices at each voltage level, i.e., whether to press the shock lever or not. This undoubtedly was a major factor accounting for the compellingness of much of Milgram's research. In contrast to the relativism and ambiguity inherent in many continuous measures (e.g., a point on a numerical scale), the discrete, observable acts comprising most of Milgram's findings lent them a quality of absoluteness, clarity, and finality that made their implications directly discernible to both lay and professional readers.

C. THE SALIENCE OF MORAL ISSUES
Another factor that has contributed to the compellingness of Milgram's work is that he often emphasized the moral aspects of the behavior under discussion. Thus, for example, he might see an observed action as representing the resolution of a conflict between two behavioral alternatives, one of which the person ought to have chosen according to moral, or at least normative, standards.
Clearly, of all of Milgram's work, the most heavily imbued with moral significance was his obedience research. In the obedience studies Milgram confronted two moral issues, the first by design and the second largely (though not solely) in response to criticisms of the ethics of the experiments. The first moral issue was embedded in the conflict situation Milgram's subjects found themselves in-a conflict between obeying legitimate authority and following the dictates of one's conscience (Milgram, 1963(Milgram, , 1974(Milgram, , 1984d. The moral dimension has been a continually salient feature of discussions of the subject's behavior in the obedience experiments, starting with Milgram himself, who described his obedient subjects as acting "in what appeared to be a shockingly immoral way" (Milgram, 1964b, p. 849). In his earliest experimental report, Milgram (1963) already pointed to the moral issue addressed by his work: "Subjects have learned from childhood that it is a fundamental breach of moral conduct to hurt another person against his will. Yet, 26 subjects abandon this tenet in following the instructions of an authority who has no special powers to enforce his commands" (p. 376). Milgram (1965a, p. 130) asked 20 college students to render a moral judgment about the experiment. After describing the concrete details of the experiment, Milgram asked "How should you perform in the experimental situation? Should you go through to the end of the experiment or should you break off at some point?" Not one subject suggested complete obedience (450 V) as appropriate, the highest stopping point mentioned being 255 V, with the median being 150V. Moral judgment also plays a salient role in the analyses of the obedience experiment by Milgram's students, Silver (1982, 1985), who see the subjects' obedience as stemming from the fact that to disobey is to reproach the authority for morally reprehensible behavior. As they put it, "Any justification they might have offered for refusing to continue would have involved an explicit or implicit condemnation of the authority" (Sabini & Silver, 1982, p. 38).
The second moral issue raised by the obedience experiments had to do with Milgram's treatment of his subjects. Beginning with Baumrind (1964), a number of writers have debated the ethics of placing subjects in the extremely stressful situation Milgram created and of deceiving them about the true purpose of the experiment (e.g., Abse, 1973;Crawford, 1972;Elms, 1982;Kelman, 1967;Milgram, 1964bMilgram, , 1973bMilgram, , 1974Miller, 1986). Milgram told an interviewer (R. I. Evans, 1976, p. 352) that he was "totally astonished" by the criticisms evoked by his obedience experiments. In an effort to dispel the claim of some critics that he was insensitive to ethical matters, in his collection of his writings Milgram (1977e) included an excerpt from his doctoral dissertation (Milgram, 1960) comparing the conformity levels of Norwegians and Frenchmen. The excerpt describes a questionnaire given to Norwegian subjects to assess their reactions to the experiment. Included in it was a question about whether the subject felt that the experiment was ethical or not. (Most thought it was neither ethical nor unethical.) Another question, similar to one he used later in the obedience experiments (Milgram, 1964b), asked subjects to indicate on a scale ranging from very glad to very sorry how they felt about being in the experiment. (The majority was either glad or very glad to have been in the conformity experiment, a finding similar to that obtained in the obedience experiments [Milgram, 1964bl.) A moral perspective was not restricted to the obedience research-it also pervaded many of Milgram's other writings. Thus, for example, he (Milgram, 1970a(Milgram, , 1970b saw withdrawal from moral and social involvement with others as one of the consequences of the stimulus overload, characteristic of urban life. A moral perspective is also evident in a "think" piece that Milgram wrote about the Kitty Genovese incident (Milgram & Hollander, 1964). In it, he points out that, in condemning the 38 nonintervening witnesses to the attack, we should not lose sight of the fact that they did not commit the murder but only failed to stop it. "It is no more than clear thinking to bear in mind the moral difference" (Milgram & Hollander, 1964, p. 602). We also find here a theme that is to recur in many places (e.g., Milgram, 1967aMilgram, , 1974, i.e., that people are not always able to translate their moral principles into action. For example, here is how Milgram (1982) describes the significance,of LatanC and Darley's research on bystander intervention: "The investigators showed that it is not just a general 'morality' that determines whether a person will help someone in trouble, but that the exact details of the situation significantly shape their responses" (p. 50).
When Milgram proposed a field experiment on television violence to the CBS network, he saw the main character in the planned stimulus program as being "confronted with a moral dilemma" (Milgram & Shotland, 1973, p. 77). Milgram even brought a moral viewpoint to phenomena one would not normally conceptualize in those terms. Thus, he saw the Asch group-pressure experiments as representing a "dilemma of truth versus conformity,'' one that the person had to resolve in a manner that was either "consistent with or in opposition to moral values" (Milgram, 1977e, p. 92). Even the innocuous act of mailing a lost letter (Milgram, 1969c) seemed to have moral overtones for Milgram.

D. THE PRIMACY OF SITUATIONAL DETERMINANTS
Earlier I noted that Milgram was an experimentalist who was able to apply the experimental method to a wide variety of phenomena. Milgram was also an experimentalist in the sense that situational factors had primacy in his thinking as potential determinants of behavior, and the understanding of a phenomenon was to be achieved through the experimental variation of relevant conditions in order to identify features of the immediate situation.that either promote or inhibit the behavior under focus. Thus, for example, in his last published statement about the obedience research (Milgram, 1984d, p. 446;1987, p. 774), he stated that "the crux of Milgram's inquiry is a set of experimental variations which examine the variables which increase or diminish obedience." More generally, he noted in another context that "social psychologists make use of experimental variation, in which one systematically alters the situation to find the causes of behavior" (Milgram & Sabini, 1979, p. 74).
Milgram was a situationist, par excellence, and his obedience research has historically provided strong ammunition for the situational emphasis of mainstream social psychology (see Blass, 1991d). The importance he placed on situational manipulations was already discernible in his doctoral dissertation (Milgram, 1960(Milgram, , 1961. In it, he set up an analog of the Asch group-pressure paradigm in Norway and France to study cross-national differences in conformity. But he used judgments of auditory stimuli, rather than the visual linejudgment task of the original Asch studies. Specifically, on each trial subjects had to indicate which of a pair of tones was longer. A unanimous majority of five was simulated through tape-recorded answers that were incorrect on designated trials. Milgram found the Norwegians more conforming than the French: The former yielded to the incorrect judgments of the bogus majority on 62% of the trials, whereas the latter conformed only 50% of the time.
In addition to establishing this basic finding, Milgram conducted a series of experimental variations. He found that the Norwegians always conformed more than the French, but also that changes in the experimental conditions affected both nationalities in similar ways. Thus, for example, both groups became less conforming when accurate responding was made more important by telling subjects that the experimental outcome would be applied to the design of safety signals on airplanes. They both became more conforming, however, in an experi-mental condition in which vocal criticism from the other "subjects" would be heard whenever a subject gave a nonconforming response.
Despite the importance Milgram placed on situational factors in understanding social behavior, he did not negate personality and individual differences, as was the case with some situationist viewpoints early in the person-situation debate (see Blass, 1977Blass, , 1984. One can readily find examples of individual dispositions receiving recognition in Milgram's writings. An early example can be found in an experiment in which he studied the effects of group pressure (rather than an authority's commands) on subjects' willingness to inflict increasingly severe shocks on a "learner" (Milgram, 1964a). The main finding was that the maximum shock elicited by two confederates playing the roles of fellow "teachers" was significantly higher than the one given by subjects in a control condition in which they were free to set the shock level on each trial. Thus, whereas only 1 subject out of 40 (2.5%) ever gave the highest (450 V) shock in the control condition, 7 out of 40 (17.5%) yielded to peer pressure to give the 450-V shock. But an almost equally salient feature of the findings were the large individual differences Milgram found in response to group pressure, with "some subjects following the group closely, others resisting effectively" (p. 140).2 A more recent example of attention to personal dispositions by Milgram comes from his writings on the psychology of photography (Milgram, 1976b(Milgram, , 1977c, in which he states that one of the greatest challenges for photography research is the delineation of the psychological makeup of the professional photographer.
More generally, Milgram saw a dispositional approach complementing his own orientation to the study of social behavior: The implicit model for [my] experimental work is that of the person influenced by social forces while often believing in his or her own independence of them. It is thus a social psychology of the reactive individual, the recipient of forces and pressures emanating from outside oneself. This represents, of course, only one side of the coin of social life, for we as individuals also initiate action out of internal needs and actively construct the social world we inhabit. But I have left to other investigators the task of examining the complementary side of our social natures. (Milgram, 1977e, p. 1)

E. THE INFLUENCE OF SOLOMON ASCH
Milgram did his graduate work in social psychology at Harvard, and the person there who had the greatest intellectual influence on him was Solomon Asch, who was there as a visiting faculty member. Milgram served as Asch's 2Judith Waters, a former student and research assistant of Milgram, notes that Milgram very briefly considered giving personality measures to subjects in the obedience experiments, but decided against it because doing so might have contaminated the expenmental results (personal communication, August 13, 1990).
teaching assistant and later also worked for him at the Institute for Advanced Study in Princeton in 1959(see R. I. Evans, 1976Tavris, 1974a, p. 77). Milgram (in Tavris, 1974a, p. 77) described Asch as "a brilliant, creative man, who possessed great philosophic depth. He is certainly the most impressive social psychologist I have known." Milgram's admiration of Asch also extended to his conformity experiment. He considered it "the decisive paradigm" for studying group effects and "as a kind of permanent intellectual jewel" (Milgram, 1977e, p. 152; see also Milgram, 1978). It is thus no surprise that in his doctoral dissertation-the cross-cultural study of conformity-Milgram ( 1960, 196 1) used an adaptation of the Asch paradigm.
More importantly, Milgram credits the Asch conformity experiment as one of the two factors that gave rise to his obedience paradigm. One factor was his general concern about authority, one that was "forced upon members of my generation, in particular upon Jews such as myself, by the atrocities of World War 11. . . . The impact of the holocaust on my own psyche energized my interest in obedience and shaped the particular form in which it was examined" (Milgram, 1977e, pp. 92-93). The other factor was his attempt to adapt the Asch conformity experiment to study something more consequential than judging lengths of lines. Here is how he put it to interviewer Tavris (1974a, p. 80) (see also Milgram, in R. I. Evans, 1976, pp. 347-348): I wondered whether groups could pressure a person into performing an act whose human import was more readily apparent, perhaps behaving aggressively toward another person, say by administering increasingly severe shocks to him. But to study the group effect you would also need an experimental control; you'd have to know how the subject performed without any group pressure. At that instant, my thought shifted, zeroing in on this experimental control. Just how far would a person go under the experimenter's orders? It was an incandescent moment, the fusion of a general idea on obedience with a specific technical procedure.
In addition to these clear-cut or explicitly stated connections to Asch, there are certain similarities between the approaches of Asch and Milgram that suggest additional instances of influence. Perhaps the most noteworthy is a shared philosophical predilection for seeing the bottle of human possibilities as half full rather than half empty. Much like Asch (e.g., Asch, 1952, p. 484), Milgram showed a preference for viewing social behavior as primarily rational, a product of the sensible weighing of available alternatives rather than of human limitation and irrationality (see, for example, Milgram & Toch, 1969, pp. 568-569).

F. INFLUENCE OF THE THEATER
The dramaturgical perspective of Goffman, suggesting that social behavior can be understood as performances staged to create a desired effect on others, has influenced a number of self-presentational perspectives in contemporary social psychology. Milgram also shows this influence when he draws on Goffman to help identify one of the "binding factors" or psychological inhibiting mechanisms that he believed prevented the subject in the obedience experiment from defying the experimenter. According to Goffman (1959), "when an individual projects a definition of the situation and thereby makes an implicit or explicit claim to be a person of a particular kind, he automatically . . . oblige[s others] to value and treat him in the manner that persons of his kind have a right to expect" (p. 13). Milgram (1974) points out that it is impossible for the subject to disobey the experimenter without at the same time violating the latter's selfdefinition as a competent authority. "Thus, the subject fears that if he breaks off, he will appear arrogant, untoward, and rude. Such emotions, although they appear small in scope alongside the violence being done to the learner, nonetheless help bind the subject into obedience" (Milgram, 1974, p. 150).
But it was not in its role as a metaphor for social life that the theater found its most pervasive influence in Milgram's work. Rather, it was in his experimentsboth in his exquisite attention to technical details and staging as well as in their intended effects on their audience (i.e., readers of his reports). In the obedience experiments, he used professional industrial engravers on his "shock generator" to ensure its perceived authenticity, and his experimenter was deliberately clothed in a gray lab coat for its ambiguous connotations about the nature of the wearer's authority, rather than white, which would signify that he was a medical technician (Meyer, 1970). Much like a director of a play, he chose his actors carefully and went through many rehearsals with them (Tavris, 1974b). A statement he once made (Milgram, 1976c, p. 24) is an apt summary of his (and many other social psychologists') approach to experimentation: "Although experiments in chemistry and physics often involve shiny equipment, flasks, and electronic gear, an experiment in social psychology smacks much more of dramaturgy or theater." For Milgram, the theater also provided a perspective from which to view a goal of experimentation and the means one sometimes has to use to achieve it. This can be seen most clearly in an exchange that Milgram had with Abse, a British playwright. Abse (1973) had written a play, The Dogs of Pavlov, which was inspired by the obedience experiments. In an introductory essay, Abse vigorously condemned Milgram's use of deception and the induction of stress, which, he believed, victimized and degraded the subject. He argued that many people "may feel that in order to demonstrate that subjects may behave like so many Eichmanns the experimenter had to act the part, to some extent, of a Himmler" (Abse, 1973, p. 29). In his rebuttal (Milgram, 1973b), which appears together with the script and Abse's introduction in one volume, Milgram expresses surprise at Abse's harsh criticisms of his use of what Milgram prefers to call "technical illusions" (rather than deception) because "as a dramatist you surely understand that illusion may serve a revelatory function, and indeed, the very possibility of theater is founded on the benign use of contrivance" (p. 39). He goes on to point out that both the playwright and the experimenter use artifice as a means toward beneficial ends-entertainment and intellectual enrichment, in the case of the theater, and the revelation of truths that are hard to get at, in the case of the experiment, i.e., that subjects were far more obedient than one would have expected. Milgram notes a further parallel between the stage and the laboratory: In both, the participants find the use of pretense acceptable. Theater goers do not feel tricked, for example, that an old man turns out to be quite young when the greasepaint is removed. In a similar vein, most subjects in the obedience experiments felt that the experience had been worthwhile, once it was explained to them.

G. STYLE, AS WELL AS SUBSTANCE
So far, my discussion of Milgram's work has focused on substance-the content of his research and writings-but style was also an integral part of his approach. His writing was invariably lucid and readable, no matter who his intended audience was-whether it was fellow psychologists reading a journal report of one of his experiments or readers of a general-circulation magazine. His judicious use of metaphor, simile, and the memorable turn of the phrase gave even his scientific journal articles a literary quality that deepened their impact on the reader. For example, in referring to the Holocaust as an antecedent of his obedience research, he said: "It has been reliably established that from 1933-1945 millions of innocent persons were systematically slaughtered on command. Gas chambers were built, death camps were guarded, daily quotas of corpses were produced with the same efficiency as the manufacture of appliances" (Milgram, 1963, p. 371). Another example, in a much lighter vein, is his description of a person's contrasting responses to two different kinds of photographs: "Examine a photograph of yourself taken away from home. . . . Count up the associations elicited by each detail of the photograph. In contrast to the garden of meanings that bloom from the snapshot taken at home, you will find this one a desert" (Milgram, 1977a, pp. 63, 65).
The literary merits of Milgram's writings have received recognition in a number of ways: His book, Obedience to Authority: An Experimental View (Milgram, 1974), was nominated for a National Book Award; parts of it were adapted for a literary magazine, Harper's (Milgram, 1973a); and his writings have been reprinted in anthologies used in writing courses (Chittenden & Kiniry, 1986;Comley, Hamilton, Klaus, Scholes, & Sommers, 1984;Eastman et al., 1988;Stubbs & Barnet, 1983).
Although much of his writing was marked by an economy of expression, Milgram's prose became rich and expansive when occasioned by the subject matter. And in fact it has been suggested that his too vivid descriptions of his subjects' emotional turmoil in the obedience studies helped convince many readers that his subjects had been mistreated (Pattullo, 1984).
One typically finished reading a piece of writing by Milgram feeling not only that one had gained some new insights about social behavior but also that the process was a thoroughly enjoyable one. Contributing to the pleasurable experience was the wry sense of humor that peppered his work throughout his career. Thus, in his very first publication, his cross-cultural study of conformity (Milgram, 1961), he noted that it was easier to simulate a majority using taperecorded voices than to use confederates because "tapes do not have to be paid by the hour and they are always available" (p. 47). And, later in his study of the cognitive maps of New Yorkers , here is how he comments on the finding that Queens residents were four times as likely to recognize a street in Manhattan than in their own borough: "Taxi drivers are reputed to fear entering Queens lest they never find their way out. And with good reason, when even the people who live in Queens are lost in their home borough compared with the sense of place they experience in Manhattan!" (p. 199). Milgram was at his best when his humor turned to sarcasm and he used it as a deadly weapon in scholarly combat with Orne and Holland (1 968). Criticizing the obedience experiments from a demand characteristics perspective, Orne and Holland argued that Milgram's subjects saw through the deception but pretended not to because they did not want to ruin the experiment. Milgram (1972) replied "Orne's suggestion that the subjects only feigned sweating, trembling, and stuttering to please the experimenter is pathetically detached from reality, equivalent to the statement that hemophiliacs bleed to keep their physicians busy" (p. 140). A biting wit enlivened his writings until the very end. In one of his last publications, Milgram (Milgram & Martin, 1985) commented on a theoretical article by wondering whether it was very interesting or simply "the commonplace raised to a high order of technical expression." And, "it is not always easy to distinguish a grand theory from one that is only grandiose" (pp. 139, 141).
Milgram was an effective communicator, ably "giving psychology away" through his writings and interviews. As he did not shy away from appearing in mass-circulation magazines, he was able to reach a wide a~d i e n c e .~ For example, an excellent concise article about his obedience research appeared in TV Guide (Milgram, 1976c) to coincide with the airing of The Tenrh Level, a television drama based on the obedience experiments. Milgram had an apprecia-3He was also an effective communicator of social psychology via the medium of film. He produced or coproduced a total of six educational films. In addition to Obedience, they are Inviration to Social Psychology, Human Aggression, Nonverbal Communication, Conformity and Independence, and the award-winning The City and the Sey. The films merit a detailed analysis, which is beyond the scope of this article. tion for the journalists' craft, and he edited a book consisting of newspaper and magazine articles on psychological topics meant for the introductory psychology course (Milgram, 1975).

Shortcomings and Flaws
So far, my overview of Milgram's work in terms of its underlying themes and characteristics has been generally favorable. But much like an aerial photograph that reveals both topographical continuities and disturbances not visible through a close-up view, an overview of the whole corpus of Milgram's work discloses not only its positive unifying qualities but also its shortcomings and flaws. One way to characterize a general deficiency in Milgram's work that a broad view reveals is that it is marked by a number of contradictions and inconsistencies, among which are the following salient discrepancies: 1. For someone so energetic in generating new ideas, Milgram's reuse of some of his material in different publications comes as a surprise. For example, we find much of the same content, with either the same or paraphrased wording, in four articles on the psychology of photography appearing in four different places (Milgram, 1976a(Milgram, , 1976b(Milgram, , 1977c(Milgram, , 1977d. As another example, most of the first chapter of the obedience book (Milgram, 1974) appeared earlier in Milgram (1967a) and the opening paragraphs were from Milgram (1963).
2. Given the generally lucid quality of Milgram's writing and the meticulous attention to detail in the design of his experiments, it is surprisingly difficult to gain a full, coherent understanding of the details of his obedience experiments due, to a large extent, to the unusual, piecemeal way they were published.
Milgram began the obedience research in 1960 and completed it in 1963. Initially, he published a series of reports on the work within a 3-year period (Milgram, 1963(Milgram, , 1964a(Milgram, , 1965a containing descriptions of varying detail of 12 experimental conditions (as well as brief references to several more). During this time, two additional articles appeared, one describing pilot procedures and findings (Milgram, 1964c) and the other (Milgram, 1964b) presenting a rebuttal of Baumrind's (1964) criticisms, which also contained new information on the well-being of his subjects, in the form of postexperimental questionnaire data and an excerpt from a brief psychiatric report (Errera, 1972). His documentary film, Obedience, also made its appearance at this time (Milgram, 1965b). In 1966, Elms and Milgram reported the results of a comparison of a sample of obedient and defiant subjects on a number of personality and other self-report measures, which Elms had administered several months after the subjects' participation in the experiments. Milgram (1967a) published an article describing his obedience research in a small-circulation British magazine. Although no new data were presented, it was here that Milgram first conveyed the importance of relinquished responsibility in the obedience process. In 1968, the most comprehensive report up to that time (Milgram, 1965c) was reprinted in the International Journal of Psychiatry together with four evaluative articles, two of which were favorable (M. Erickson, 1968;Etzioni, 1968) and two which were critical of the obedience work (Masserman, 1968;Orne & Holland, 1968). The same issue also contained a brief reply to his critics by Milgram (1968), most of which was directed at Orne and Holland (1968), the longer of the two critical articles. The latter had argued that Milgram's subjects probably saw through the deception. Later, Milgram (1972) wrote a much longer rebuttal of Orne and Holland, which included newly reported postexperimental data in support of Milgram's claim that most subjects believed the shocks were genuine and very painful. A full description of the research program did not appear until 1974 (Milgram, 1974). This book presented 19 variations in similar detail (i.e., in addition to the verbal descriptions, frequency distributions of break-off points for each condition appeared in tabular form). Nine of these conditions were described for the first time. Ten had appeared in earlier reports, but no cross-referencing-which would have enabled the reader to integrate all the findings and publications-was provided.
Reading the various reports of the obedience experiments also leaves one with a lack of sense of closure. First, this is due to the brief mention of a couple of conditions without a precise quantitative report of their outcomes, i.e., a condition in which the experimenter is completely absent and tape-recorded instructions are used and another in which the experimenter is absent during the first part of the experiment, giving his instructions by phone, but then reappears when the subject refuses to continue (Milgram,196%). Second, at the end of one of his reports (Milgram, 1965c, p. 71) Milgram lists a number of variations that were to be reported elsewhere. All do appear in Milgram (1974) except for one, referred to as an experiment that "concerned the personal relationship between victim and subject" (Milgram, 1965c, p. 71). This description does not seem to match any of the variations reported later in Milgram (1974) nor in any of Milgram's other publications. (See Blass, 1991d, for a review of the obedience research.) Although "a foolish consistency" may be "the hobgoblin of little minds," as Emerson once said, one cannot help but wonder about the following additional contradictions in Milgram's work: 3. Why are self-reports considered credible some of the time but not other times? In his reply to Orne and Holland's (1968) criticisms of the internal validity of the obedience studies, Milgram (1972 , Table 2-3) presented the results of a follow-up questionnaire given to subjects a year after their participation, which revealed that only 2.4% of the subjects were sure that the learner was not really getting shocked. But he was not willing to discard the data from these kinds of subjects because their stated disbelief may well have been an after-the-fact rationalization to justify their obedience. In another context (Milgram, 1964b)-in attempting to justify the ethics of the obedience research-he also draws on questionnaire responses of subjects for supportive data. Milgram found 83.7% of his subjects stating that they were glad or very glad to have participated, 15.1 % were neither sorry nor glad, and only 1.3% had misgivings about it. Initially (Milgram, 1964b) he noted that "such findings are to be interpreted cautiously, but they cannot be disregarded" (p. 849). Later, however, caution seemed to have been cast aside, with Milgram (1977h) dismissing the criticism that subjects' positive reactions may simply have been a product of cognitive dissonance. In fact, in his later writings, subjects' positive evaluations of their experience were made to bear the main burden for the ethical justification of the obedience studies: "The central moral justification for allowing a procedure of the sort used in my experiment is that it is judged acceptable by those who have taken part in it. Moreover, it was the salience of this fact throughout that constituted the chief moral warrant for the continuation of the experiments" (Milgram, 1973b, pp. 40-41; see also Milgram, 19771.1, pp. 21-22).4 4. Why did a researcher, who for most of his career had, by word and deed, stressed the importance of studying behavior directly (as opposed to what people say they will do), study cognitive maps without also demonstrating their links to actual behavior? Milgram conducted three systematic studies of the mental maps of the residents of two major cities, New York City and Paris. The primary focus of the first of two New York City studies  was to determine the recognizability of different parts of the city by its inhabitants. To achieve this goal, Milgram and his students systematically sampled and photographed over 150 scenes throughout the five boroughs of the city. A group of New Yorkers, proportionately drawn from all the five boroughs, were subsequently asked to identify the locations depicted in the pictures. The Paris study (Milgram, 1984a;Milgram & Jodelet, 1976 used a number of other techniques besides scene recognition-e.g., having subjects draw subjective maps of the city-in order to tap Parisians' cognitive representations of their city. Milgram also conducted a second study of New Yorkers' images of their city using hand-drawn maps and responses to a questionnaire published in New York magazine (Duncan, 1977;"Mental Maps," 1975;Milgram, 1984a). Although the studies were imaginative 4Judith Waters (personal communication, August 13, 1990) adds: "I read all of the responses when I was preparing to illustrate the book [Milgram, 19741. Many of the comments sound very defensive or as if they (the respondents) were trying to impress him (and Yale). I remember one, in particular, who said that participating in the experiment was the most important event of his life. One wonders about the rest of that existence!" and many of the results fascinating, the connections to actual behavior, though assumed, were never dem~nstrated.~

IV. Enduring Contributions
Although Milgram's work did have some shortcomings, they were clearly overshadowed by his enduring contributions to social psychology, as well as to some other disciplines, to which we now turn. Specifically, I will assess the impact of each of his major research foci in turn.

A. OBEDIENCE TO AUTHORITY
Within social psychology, and perhaps even psychology as a whole, it would be hard to match the durability and breadth of interest evoked by the obedience research. From the very beginning, it caught the attention of both psychologists and the press (e.g., Baumrind, 1964;"Could We Be," 1964;Kaufmann, 1967;Sullivan, 1963) and, to the present day, continues to stimulate research and analysis (Blass, 1990a(Blass, , 1990b(Blass, , 1990c(Blass, , 1991a(Blass, , 1991b(Blass, , 1991dBlass & Krackow, 1991;Gibson, 1991;Meeus & Raaijmakers, 1987;Miller, 1986;Nissani, 1990;Schurz, 1985) and influence-or, at least, figure prominently in-theorizing about obedience-related phenomena (Billig, 1990;Haritos-Fatouros, 1988;Kelman & Hamilton, 1989;Staub, 1989). A significant portion of Mixon's (1989) book is devoted to refuting Milgram's view that his subjects' actions represented destructive obedience and to presenting an alternate conceptualization, building on earlier writings (e.g., Mixon, 1971Mixon, , 1976. Not only does the work continue to be cited in most introductory texts (see Gorenflo & McConnell, 1991) and virtually all social psychology texts, but it also appears in some junior and senior high school curricula on the Holocaust (Strom & Parsons, 1978; Teaching About the Holocaust, 1985) and in one aimed 5John Sabini, a former student and research assistant of Milgram, provides a perspective that can help diminish the inconsistency between the cognitive map studies and Milgram's other work: "It is true that Milgram never demonstrated connections to behavior [in the cognitive map studies]. But as you . . . point out early in the paper, Milgram wasn't a behaviorist. He was interested in capturing aspects of experience. It is also true, of course, that his favored strategy was to find some dramatic behavior to externalize that experience. But in the case of mental maps I think he thought that in themselves they captured something of the subjects' experience of the cities. The spirit of his thinking about mental maps seems clearest to me not in his articles about them, but in his writing about photography" (personal communication, August 20, 1990). more generally at sensitizing students about the dangers of blind obedience (Bushman, 1985).
Unmatched, too, perhaps, are the intense opinions, pro and con, that the obedience research has generated. For example, Roger Brown (1974) stated that the "series of experiments . . . probably constitutes the most important social psychological research done in this generation" ( p. 42), and Herrnstein (1974) regarded them as "among the most original in all of social psychology" (p. 82). In a similar vein, Eysenck (1974) noted that "Many people believe that an experimental social psychology is either impossible, sterile, or useless. That this belief is mistaken has been obvious for quite some time. . . . The important contributions of Stanley Milgram during the early sixties present us with another outstanding example of the possibilities inherent in the use of the experimental method in this field" (p. 21). More recently, Ross (1988) articulated the importance of the obedience research as follows: "Perhaps more than any other empirical contributions in the history of social science, they have become part of our society's shared intellectual legacy-that small body of historical incidents, biblical parables, and classic literature that serious thinkers feel free to draw on when they debate about human nature or contemplate human history" (p. 101).
On the other hand, Wrightsman (1974) gave this assessment of the obedience work: "When I consider that the initial study was a demonstration and not even an experiment, that the research program lacked any initial theory or tests of significance, and that many of its conclusions are subject to alternative explanations, I am saddened that it is the obedience study that will go down in history as reflecting the 1960s in social psychological research" ( p. 804). Kohlberg (1974) also took a critical stand toward the obedience experiments, though on different grounds: "Milgram himself was unwittingly the moral victim of the 'authority of science,' just as his subjects were. Serving the authority of science under the banner of 'objectivity,' he, himself, inflicted pain on others for the greater social welfare. . . . Milgram failed to treat his subjects as persons, the basic meaning of the moral attitude, just as most of his subjects failed to treat the victim as a person" (p. 607).
The most severe-and surprisingly strident-criticism of the obedience work that I have encountered came from Bruno Bettelheim (in Askenasy, 1978, p. It is a testament to the broader human significance of the obedience work that disciplines outside of psychology, e.g., economics (Akerlof, 1991), education (Atlas, 1985), sociology (Damico, 1982), political science (Helm & Morelli, 1979, and philosophy (Patten, 1977), as well as political and social commentators (e.g., Giffin, 1983;Karnow, 1971;Krauthammer, 1985;Shnayerson, 1973;S. J. Singer, 1991), have found relevance in it. By 1981, the first journal report on the obedience research (Milgram, 1963) had become a "citation classic," having been cited over 255 times since 1966 (Milgram, 1981a). The continuing interest in the obedience studies has transcended national boundaries, as evidenced by the fact that Obedience to Authority: An Experimental View (Milgram, 1974) has been translated into 1 1 languages (with multiple editions appearing in some cases), i.e., German (three editions), French (two editions), Japanese (two editions), Dutch, Danish, Italian, Spanish (two editions), Swedish, Portuguese (published in Brazil), Indonesian, and Serbo-Croatian (published in Yugoslavia). There is also a British edition of the book. Milgram continued to write and speak about the topic long after he had completed the research (CBS News, 1979;Milgram, 1979Milgram, , 1983Milgram, , 1984dMilgram, , 1987Muson, 1978).
Not surprisingly, in a recent article on "psychological literacy" (Boneau, I990), Milgram's obedience experiment is listed among psychology's "top 100" terms and concepts. Nor should it be surprising that one of the exhibits comprising an innovative psychology exhibition, which opened at the Ontario Science Center in Toronto in March, 1991, is about the obedience research ("Exhibition Explores," 1991). (In it, visitors entering a tiled passageway are instructed by signs to walk on the black squares only. Almost everyone obeys. Afterward, they get an introduction to the obedience experiments, which includes seeing the original "shock" generator used by Milgram.) But it is quite remarkable that professions as diverse as marketing (Armstrong,I975), accounting (Dirsmith, 1983), nursing (Redfearn, 1982), and management (Laurent, 1987) can derive practical lessons from the obedience studies for their respective practitioners. Even more remarkable is a recent account of a social psychologist's trial testimony in which the lessons derived from the obedience research (as well as some other classic social-psychological studies, such as those on group polarization, bystander intervention, and deindividuation) actually saved lives. In two trials in South Africa involving a total of 13 defendants accused of murder during mob actions, the courts accepted obedience to authority and several other social psychological phenomena as extenuating factors, with the end result that nine of the accused were saved from the death sentence (Colman, 1991).
The obedience work dramatically demonstrated that people are much more likely to obey the orders-perhaps even the destructive orders-of a legitimate authority than was thought.6 Milgram's (1965c, p. 74) suggestion that this involves allowing the authority to define reality for us, "the control of [our] 6Though this remains an enduring lesson, there is some evidence (Kaufmann & Kooman, 1967;Mixon, 1971) suggesting that it is in need of some qualification.
perceptions," is both thought-provoking and disturbing. Although, as I will show, the exact parallels between the behavior of his subjects and those of the Nazis and their collaborators remain open to debate, there is no question that the obedience experiments have promoted a greater awareness of the Holocaust and efforts to understand its determinants. I hope that such consciousness raising can serve a preventive function. As a Holocaust survivor, I am especially appreciative of the potential value of the obedience experiments in this regard.
The ethical controversy surrounding the obedience experiments has left an indelible mark on experimentation with human subjects. It, together with a handful of other controversial studies, led to the formulation of the American Psychological Association's ethical principles for research with human subjects and to the federally mandated creation of Institutional Review Boards for the protection of human subjects. Although psychologists have disagreed about the necessity of such regulations, there is currently a greater sensitivity to the wellbeing of the research participant than was the case in the past, as evidenced, for example, by the quadrupling of the number of studies reporting debriefings from 1964 to 1980 in the Journal of Abnormal and Social Psychology and Journal of Personality and Social Psychology (Ullman & Jackson, 1982). As Miller (1986) aptly concluded after reviewing the ethical debate surrounding the obedience studies, "To say that we are better off for having consjdered the ethical dimension of the obedience paradigm may sound trite, but it must certainly be the case. Among the beneficiaries, doubtless, are legions of research subjects who have been treated in a more humane and informed manner by researchers who, in the past two decades, have been enlightened by studying the issues reviewed here" (p. 138). The obedience studies are often used to illustrate ethically problematic research in introductory, social, and methods texts, but what is often overlooked is that Milgram (1963) undertook procedures "to assure that the subject would leave the laboratory in a state of well-being. A friendly reconciliation was arranged between the subject and the victim, and an effort was made to reduce any tensions that arose as a result of the experiment" (p. 374). In this regard, it is interesting to note that Harris (1988) credits Milgram (1964b) with the first published use of the term "debriefing" in an experimental context.

I . Impact on Social Psychological Research and Conceptualizations: Variety and Numbers
We now turn to a more focused look at the discernible influences of the obedience paradigm on contemporary social psychological research and thinking.
Over the years, the obedience studies have been used by social psychologists to crystallize concepts, to clarify phenomena, and to marshal1 support for viewpoints on a variety of issues. Thus, when Freedman (1969) argued against role playing as an alternative to the standard laboratory experiment, he drew on the oft-reported gap between predictions and findings in the obedience experiment as major support. (Ironically, it later turned out that the strongest evidence supporting the research viability of role playing came from three role-playing variations of the obedience experiment that obtained levels of obedience comparable to Milgram's findings [Geller, 1975[Geller, , 1978Mixon, 1971; O'Leary, Willis, & Tomich, 1970; see also Albrecht, 19731.) When a number of critics declared a crisis in social psychology during the late 1960s and early 1970s and accused it of triviality and irrelevance, defenders of mainstream social psychology would invariably counter by citing the obedience work as a prime example of substance and social relevance. And when Aronson and Carlsmith (1968) introduced the distinction between mundane and experimental realism, didn't their choice of the obedience experiment as an illustration of the latter help make what they meant vividly clear? When Gergen (1973) argued for the historical and cultural boundedness of many social psychological phenomena, he used the obedience research as one of his supporting pieces of evidence, suggesting that the findings were dependent on prevailing attitudes about authority.
The obedience paradigm also came into play in relation to another of Gergen's (1973) contentions, i.e., the operation of enlightenment effects. Gergen argued that "sophistication as to psychological principles liberates one from their behavioral implications" ( p. 3 13). To provide a concrete test of whether or not obtaining psychological knowledge can affect one's behavior, as Gergen claimed, Shelton (1982) conducted an ingenious study in which she first taught her subjects about the basic obedience experiment and its outcome. Then she had them serve as experimenters in a similar "learning" experiment in which they had to supervise another subject who was supposed to teach a verbal-learning task to a learner by means of increasing voltages of shock on each subsequent error. Although both teacher and learner were confederates, the experimenter-subject was told that only the latter was. As the voltages of "shock" and the learner's expression of pain increased, the teacher, on the verge of crying and complaining of stomach pains, begged the experimenter-subject to stop the experiment. In spite of this, out of 24 subjects who began the experiment, 22 went to the end, commanding the teacher to continue to the maximum shock-450 V. Apparently, the subject could not see that his or her obedience to the investigator was similar to the teacher's obedience to him or her.
The obedience research played an evidential role in Ross's (1977) introduction of the fundamental attribution error, i.e., the tendency for perceivers to overemphasize dispositions and underemphasize situational determinants in explaining another person's behavior. Ross argued that the obedience research and Asch's conformity studies were the successes they were among psychologists "because they demonstrated control by a situational factor that the reader had previously assumed to be too weak to exert such control" (p. 187).
The obedience experiment also played a role, though a lesser one, in another type of social perceptual phenomenon-stereotyping. One of the earliest studies providing evidence for the beauty stereotype was one by Miller, Gillen, Schenker, and Radlove (1974). In that study, subjects were given photographs of individuals differing in attractiveness and were asked to indicate the maximum shock these individuals would have given the learner had they been in the obedience experiment. Miller and colleagues found that the less attractive persons were seen as giving higher shocks than the more attractive ones.
More recently, I have been engaged in a research program exploring perceivers' attributions about the behavior of the interactants in the obedience scenario, using an edited version of Milgram's (1965b) documentary film, Obedience (Blass, 1990a(Blass, , 1990b(Blass, , 1991b. For example, one study (Blass, 1990b) attempted to replicate a finding by Tyler and Devinitz (1981). In their experiment, subjects read about a theft from a dormitory room that was described as either occurring quite often (to 1 out of 20 students) or very infrequently (to only 1 out of 200 students). Tyler and Devinitz also manipulated outcome severity by varying the value of the stolen item (either $5 or a stereo worth $300). They found that the theft victim was held more responsible for the rarely occurring theft than for the more frequent one, but that severity of outcome had no effect on responsibility attributions. My study was a conceptual replication of the Qler-Devinitz study aimed at determining if frequency rather than severity would still influence attributions even in a highly involving task. I showed subjects a short version of Obedience, which ends with a participant continuing to increase the voltage on the shock generator. The film was stopped right after the participant gives the 180-V shock. Outcome frequency and severity were varied in a factorial design via differential information given to subjects. Some subjects were informed that the subject in the film went no further, whereas others were told that he continued to the final shock (mild versus severe outcome). Outcome frequency was varied by informing subjects that he was either 1 of 15 or of 65 out of a total of 100 subjects who did this. Confirming Qler and Devinitz's results, I found that outcome frequency, but not severity, had a significant effect on responsibility attributions.
Another recent instance of the influence of the obedience paradigm can be seen in Cialdini's (1985Cialdini's ( , 1987 organizing framework for the various sources of interpersonal influence. Drawing on the findings of experimental social psychology and the tactics used by compliance professionals, such as car salesmen, Cialdini formulated the following six principles of compliance: reciprocation, commitment and consistency, social proof, liking, scarcity, and authority. His prime support for the power of the authority principle is, of course, the obedience studies (see Cialdini, 1985, pp. 175-180).
We have seen the variety of ways the obedience paradigm has been implicated in social psychological research and thought. But ultimately the prime indicators of the impact of a paradigm are, first, the extent to which it has stimulated further research involving the paradigm, and, second, the increment in knowledge of the phenomenon provided by that research. Over the years, a number of investigators have conducted replications of the Milgram obedience experiment. By my count, there are about 40 reports of replications in the literature other than Milgram's own studies. This figure includes both methodological and conceptual replications as well as role-playing versions of the obedience experiment. Although many of the reports involve more than one experimental condition, this is still a rather modest yield considering the amount of attention the obedience studies have received over the years, and in comparison to the number of studies generated by some other social psychological paradigms, such as the risky shift (Cartwright, 1973). Undoubtedly, the advent of federal regulations and the American Psychological Association's code of ethics on research with human subjects in the early and mid-1970s effectively curtailed the further use of the obedience paradigm in the United States so that, ironically, within a year or two of the publication of Obedience to Authority: An Experimental View (Milgram, 1974), such research by Americans had ground to a halt. It is thus no coincidence that the most recent replications have been conducted in countries other than the United States, i.e., Scotland (Burley & McGuinness, 1977), Jordan (Shanab & Yahya, 1977, Spain (Miranda, Caballero, Gomez, & Zamorano, 1981), Canada (Shelton, 1982), Austria (Schurz, 1983, andHolland (Meeus &Raaijmakers, 1987).

. Impact on Advancing the Understanding of the
More important than their sheer numbers is the advance in our knowledge of the dynamics of obedience to authority, beyond Milgram's work, that the replications have provided, to which we now turn. Specifically, I will review the accumulated evidence pertaining to three salient questions relating to the Milgram obedience research: Is the subjects' behavior in the Milgram-type experiment a manifestation of obedience or of aggression? What is the current status of the "agentic-state" ~o n c e p t ?~ Did Milgram demonstrate the operation of destructive obedience?
One of the most clear-cut and consistent findings in the obedience literature is that when the subject is allowed to choose the shock level on each trial, rather than being required by the authority to increase it continuously, the subject typically remains at the lower end of the voltage scale and only rarely does a subject use the highest voltage switch. In

Dynamics of Obedience a. Is I t Obedience or Aggression?
7Most of the material that follows pertaining to these two questions was presented at the symposium on Milgram's contributions to social psychology held at the 1990 APA meetings in Boston, Massachusetts (Blass, 1990~). such a variant (Milgram, 1974, Expt. ll), Milgram found that the average highest shock level chosen by the subjects was 5.5 (on the 30-step shock scale), and only 1 subject out of 40 used the highest shock lever. This is to be contrasted with the appropriate comparison baseline condition (Expt. 5), in which the average maximum shock administered was 24.55, with 65% of the subjects being fully obedient (see Milgram, 1974, Table 3, p. 60).
Similar results have been found by others. In an obedience study conducted by Mantell (1971) in Munich, Germany, he found that only 7% of his subjects in a self-decision condition gave the highest shock, compared to an 85% obedience rate in his baseline condition, a voice-feedback variant. In Jordan, Shanab and Yahya (1977) found that only 16% of the subjects (public school children) in a self-decision condition delivered the highest shock, a figure significantly lower than the obedience rate in a standard condition (73%). In an experiment using the same procedures with an adult sample (Jordanian college students), the analogous figures were 12.5 and 62.5%, respectively (Shanab & Yahya, 1978). It should be noted that the shock levels reported by Shanab and Yahya in their selfdecision conditions are higher than those found by Milgram (1974, Expt . 1 1) and Mantell (1971). This may have been due to a feature of the self-decision condition unique to Yahya (1977, 1978), i.e., although subjects were free not to shock at all on any given trial, once they did give a shock, it had to be one step higher on each subsequent error.
Another study (Kilham & Mann, 1974), conducted in Australia, also found very low shock levels attained in a self-decision condition. The mean shock level was 3.12, a figure that was significantly lower than the ones in their more standard conditions. A doctoral dissertation by Shalala ( 1974) comprised a rather complicated obedience experiment. Subjects were military personnel at Fort Knox, Kentucky, and the experiment involved seven different conditions. One of them was a selfdecision condition (SD,), which appears to be somewhat comparable to the ones in the experiments described so far. In this condition, the subjects, low-ranking Army tank crewmen, were told by the experimenter, a lieutenant colonel, to "use their own judgments as to when, if, and how much punishment" the learner was to get for making mistakes (p. 50). No subject in this condition gave the highest shock on the scale, compared to a 30% obedience rate in a voice-feedback condition.
The distinction between a condition in which a subject is free to choose the shock intensity with which to "punish" a "learner" and one in which a subject is commanded to increase the intensity by one step on each subsequent error is fundamentally important, for two reasons. One is that this constitutes the methodological difference between the Buss aggression paradigm and the Milgram obedience paradigm. Although both Buss and Milgram independently developed a shock machine that allowed for shocks of increasing voltage levels to be administered, the form of their typical use has varied. Buss and other aggression researchers have allowed the subject to choose from among several shock intensities on each punishment trial and studied the average shock levels (and sometimes durations) chosen as a function of different antecedent conditions (e.g., frustration, presence of cues suggesting violence). Intrinsic to the Milgram obedience format, however, is the incremental nature of the procedure, so that the focus of interest is on the average break-off point or the number of subjects proceeding to the highest possible shock level. The large difference between a self-decision procedure and the standard incremental format in punishment levels typically administered also has theoretical importance. A possible contending explanation, noted by Milgram, for the fact that his subjects were willing to use very high shock intensities to punish a learner is that subjects' aggressive tendencies rather than the experimenter's orders were causal (Milgram, 1974, p. 70). That is, according to this view, the experiment merely provides the opportunity for the expression of hostile impulses that people normally have to hold in check. Thus, from this perspective, the apparently severe punishments subjects were willing to inflict on another person would not be a manifestation of the power of an authority to exact obedience, but rather a demonstration of what happens when a person's aggressive instincts are given free rein. The low shock intensities delivered by subjects in Milgram's self-decision experiment (Expt. 1 l), and similar findings by the other researchers, which I have described, argue strongly against this alternative explanation. In a self-decision condition, inflicting severe punishments was as acceptable as when subjects were ordered to increase the shock level for each subsequent punishment. Yet in the former, subjects typically restricted their punishments to the low end of the voltage scale. As Milgram (1974) put it, "Insofar as the experiments tell us something about human nature, the revelation on how men act toward others when they are on their own is here. Whatever leads to shocking the victim at the highest level cannot be explained by autonomously generated aggression but needs to be explained by the transformation of behavior that comes about through obedience to orders" (p. 72). It should be noted, parenthetically, that these findings do not completely rule out the operation of individual differences in hostility or aggressiveness. Although subjects in self-decision conditions typically stay at the lower ends of the scale, there is still some variability there that could be due to dispositional differences in aggressiveness.
The results of three other variations conducted by Milgram (1974) also speak against the primary determinative role of aggressive tendencies in his experiments. One was Experiment 12, in which the experimenter calls a halt to the experiment but the "learner" wants it to continue, claiming that "it would be an affront to his manliness" not to finish it. A second was Experiment 13, in which the experimenter gets called away, after giving instructions to the subject and to a confederate playing the role of another subject. He tells them to go ahead with the experiment until the learner learns all the word pairs correctly but says nothing about which shock levels they should use. After the experimenter leaves, the confederate apparently thinks up the idea of increasing the shock on each subsequent error and orders the subject to follow this procedure. In the third experiment (Expt. 14), the authority ends up as the victim, with an ordinary person giving orders to the subject to shock him. This reversal comes about as follows: After the experimenter describes the punishments to be used, the learner (a confederate) says he is afraid of shocks and is willing to participate only after he sees someone else-for example, the experimenter-go through the same procedures. Because of the difficulty he has had in obtaining subjects, the experimenter agrees, provided that the learner serve in the experiment right afterward. The experiment proceeds with the subject administering the shocks. At 150 V, the experimenter demands to be let out but the confederate insists that they continue. Both in this experiment and in the one in which the learner wants to continue but the experimenter says to stop (Expt. 12), no subject continued giving shocks after the experimenter called a halt to the proceedings, and in Experiment 13, in which a peer was giving the orders, only 4 out of 20 subjects ended up giving the highest shock level. Milgram summarizes the significance of these three experiments as follows: "Those who argue that aggressive motives or sadistic instincts are unleashed when the command to hurt another person is given must take account of the subjects' adamant refusal to go on in these experiments. It is not what subjects do but for whom they are doing it that counts" (p. 104). So the evidence is solidly supportive of Milgram's view that the power of the authority, rather than aggressive tendencies, was the primary determinant of the punishing behavior of his subjects. b

. What Is the Current Status of the Agentic-State Concept?
When we turn to the literature pertaining to the agentic-state concept, however, the evidence is clearly not as supportive. Milgram (1974) had argued that obedience involves divesting oneself of the responsibility for one's actions and relinquishing it to the authority in charge, and he referred to this process as a shift from a state of autonomy to an agentic state. It was seen by him as the process that enables a person to act destructively, without the usual moral constraints on his or her actions.
Although logically compelling, the evidence supporting the shift-in-responsibility process has been weak or contradictory. Milgram's (1974, pp. 203-204) own evidence comes from postexperimental information gathered from subjects in his proximity series (Expts. 1 through 4 in Milgram, 1974). He asked each participant to divide responsibility among the experimenter, the victim, and himself for giving the victim "shocks against his will." In line with his agenticstate notion, defiant subjects ascribed a higher percentage of the responsibility to themselves (48.4%) than did obedient subjects (36.3%). However, this finding is weakened by the fact that obedients (38.4%) did not relinquish any more responsibility to the experimenter than did defiants (38.8%). Obedients did, however, place twice as much responsibility (25.3%) on the victim than did defiants (12.8%). Although this is an interesting finding given what we know historically about the dehumanization of victims of mass violence, it is not intrinsic to the agentic-shift process. Mantel1 and Panzarella (1976) had three experimental conditions in the previously mentioned obedience experiment conducted in Munich, Germany: a standard, baseline condition, a "delegitimizing model" condition, and a selfdecision condition. Although obedience varied as a function of condition, how much responsibility the subject assigned to himself was not significantly related to amount of obedience. In an experiment by Tilker (1970), subjects were made observers of a Milgram-type experiment and the amount of responsibility for the welfare of the learner was varied. Amount of assigned responsibility did not have a significant overall effect on how much the observer protested the teacher's actions. However, all five subjects in a condition in which they were given total responsibility and could see and hear the victim physically interfered with the experiment. Also contradictory is the previously mentioned experiment by Kilham and Mann (1974) conducted in Australia. On the one hand, consistent with a shift-in-responsibility process, was their finding that obedience was significantly reduced when the subject had only the indirect role of transmitting the experimenter's orders rather than actually pressing the shock levers. Milgram (1974, Expt. 18) had obtained a similar result. On the other hand, Kilham and Mann's experiment had an unusual feature that should have led to increased obedience according to Milgram's agentic-state conceptualization. Their fourth and final prod to be used on reluctant subjects was "I'm taking full responsibility, so you have no option but to continue with the experiment" (p. 699). (It should be noted that although Milgram's experimenter apparently occasionally accepted responsibility explicitly [see Milgram, 1965bMilgram, , 1965cMilgram, , 1974, this was not one of Milgram's prods nor otherwise a planned feature of the experimental script.) Despite this explicit reference to responsibility-taking by the experimenter, the percentage of subjects who were fully obedient (28%) is the lowest reported in the literature for a standard, baseline condition. A final study, a recent obedience experiment by Schurz (1985) in Austria, is supportive of Milgram. Instead of using electric shock, her subjects gave what they thought was increasingly painful ultrasound to a learner. At the highest points on the scale, this form of stimulation was supposedly capable of causing injury to the skin. Supporting Milgram's view, she found that disobedient subjects were more willing to take responsibility for their actions than those who were fully obedient.
It remains a challenge for future thought and analysis to reconcile this set of findings with Milgram's agentic-state model. One direction this reconciliation might take is the possibility that only some subjects need to relinquish responsi-bility in order to be obedient, a possibility suggested by the findings of an obedience experiment conducted in Italy by Ancona and Pareyson (1968).* In their experiment, subjects were instructed to give increasingly painful shocks via a 22-step 330-V shock generator. The victim, played by a professional film actor, was both visible and audible and responded with increasingly intense cries of pain and body movements as the subject increased the voltage, and by 240V the victim was begging "Stop now!" Altogether, 34 out of 40 subjects (85%) were fully obedient. On the basis of the subjects' answers to a postexperimental questionnaire and the clinical examination of the postexperimental interviews by three judges, Ancona and Pareyson were able to classify the obedient subjects into two groups of 17 each. They report that subjects in the first group participated because they trusted the authority behind the experiment, were preoccupied with the victim's welfare, and accepted responsibility for any negative consequences of their own actions. On the other hand, the second group of obedients participated in the experiment out of obedience to an order, downplayed the suffering of the victim, focused on their own reactions rather than the victim's, and denied any responsibility for their own actions. Although some caution is called for because clinical judgment was involved in distinguishing the two groups, these findings are important. They suggest that shedding responsibility is a concomitant of obedience only for some people (i.e., not everybody has to enter into an agentic state in order to be obedient), which might explain the conflicting results pertaining to responsibility-taking and to relinquishing that I just reviewed.
Unquestionably, one of the most important factors behind the continuing and widespread interest in the obedience work is the disturbing lesson Milgram derives from it about human nature, i.e., just how easily normal individuals can be made to carry out inhumane commands. In particular is the implication that anybody could have acted like the Nazis and willingly participated in the destruction of European Jewry during the Second World War.
The potential relevance of his work to the Holocaust was a continuing theme in Milgram's writings, beginning with the opening paragraph of the first report of his findings (Milgram, 1963). Although Milgram was certainly sensitive to the differences between the conditions in his laboratory and those during the Nazi era ("Is a match flame comparable to the Chicago fire of 1898?," Milgram, 1974, p. 175), he felt that both events shared a common psychological process: "In the laboratory, through a set of simple manipulations, ordinary people no longer perceived themselves as a responsible part of the causal chain leading to action against a person. . . . What we find in common among soldier, party function-

c. Did Milgram Demonstrate Destructive Obedience?
*My thanks to Alfonso Campolattaro for putting much time and effort into translating this article from the Italian. ary, and obedient subject is the same limitless capacity to yield to authority and the use of identical mental mechanisms to reduce the strain of acting against a helpless victim" (Milgram, 1974, pp. 175-176). And, in 1979, appearing on the TV news program, "Sixty Minutes" (CBS News, 1979), Milgram stated, on the basis of his findings, that "if a system of death camps were set up in the United States of the sort we had seen in Nazi Germany, one would be able to find sufficient personnel for those camps in any medium-sized American town" ( pp. Milgram's analysis has been adopted or at least gets sympathetic treatment not only by many introductory-level texts, but also by more advanced and scholarly works, e.g., Sabini and Silver (1980), Berger (1983), Miller (1986, and Hirsch (1988). A book called Are We All Nazis? by Askenasy (1978) answers that question largely on the basis of Milgram's work. More recently, in Kelman and Hamilton's (1989) book, Crimes of Obedience, the obedience work is used as the model for the difficulty people have in defying authority.
The validity of Milgram's assertion that there is a parallel between the behavior of his laboratory subjects and that of the Nazis under Hitler rests on the assumption that obedient subjects in his obedience experiments interpreted the situation the way he had intended it. But is this assumption correct? More specifically, there are two related but separate questions that bear on this issue. First, did subjects believe that the shocks were real? Second, even if they believed that the shocks were genuine, did they also believe that they were harming the victim? In other words, did their actions represent "destructive obedience" as Milgram (1963, p. 371) contended, or merely that they were inflicting pain but not permanent injury?
The question about the believability of Milgram's procedures was first raised by Orne and Holland (Holland, 1967;Orne & Holland, 1968), who argued that the demand characteristics of the setting-i.e., the incongruity of a calm experimenter prodding the subject to go on despite the protests of the learner-revealed to the subject the make-believe nature of the proceedings. If this interpretation of subjects' behavior in the Milgram experiments is correct, it calls for a differing view regarding the possible parallels between them and Nazi Germany. As Holland put it, "Milgram's subjects were obedient, but they were obedient only in that they pushed levers. Death-camp executioners were obedient, but they were obedient in that they pushed levers knowing that in doing so they were committing murder. No real relationship may be drawn between these two cases unless it can be established that Milgram's subjects 'knew' that they were shocking the learner as they had been told. This has not been established" (Holland, 1967, p. 71). Milgram (1968Milgram ( , 1972 presented an effective rebuttal of Orne and Holland's critique, drawing on postexperimental data (Milgram, 1972) to show that most of his subjects believed that the shocks were real and extremely painful. It should 7-8).
also be noted that if we examine the replications of the obedience studies that have been conducted outside of North America-in Jordan (Shanab & Yahya, 1977, West Germany (Mantell, 1971), Spain (Miranda et al., 1981), Italy (Ancona & Pareyson, 19681, Austria (Schurz, 1985), and Australia (Kilham & Mann, 1974)-we can note that, in all cases except the Australian replication, the obedience rates were as high or higher than the ones reported by Milgram (1974) for comparable conditions. Subjects in these countries were probably less familiar than North Americans with psychological experimentation and therefore less likely to have the kind of problem-solving set postulated for subjects in psychology experiments by Orne and Holland. Their high rates of obedience speak strongly in favor of the credibility of the obedience experiment for most subjects.
Even if the shocks in the obedience experiments were perceived as real, and therefore the learner's protests genuine, can we also assume that subjects obeyed believing that their actions were destructive, i.e., injuring or killing the victim? The fit between the behavior of subjects in an obedience experiment and the murderous actions of a Nazi stormtrooper is contingent on the assumption that obedient laboratory subjects saw their actions as potentially deadly.
This assumption has been questioned by several writers. Mixon (1971Mixon ( , 1976Mixon ( , 1989 has argued that, although subjects might be willing to cause pain, they would not harm another individual. If subjects were certain that the learner was being harmed, they would not obey the experimenter. According to Mixon, subjects obey the experimenter, knowing that in a scientific experiment the safety of participants is assured and trusting the experimenter when he says that although the shocks may be painful, they are not dangerous. Disobedience takes place, according to Mixon, when the learner's protests make the subject question the experimenter's reassurances. Mixon's view of the obedient subjects in a Milgram experiment is therefore a more benign one, explaining their behavior without recourse to any assumptions about their destructiveness. The empirical evidence in support of Mixon's position comes from a nonactive role-playing procedure. In it, Mixon (1971) read subjects a description of the procedure in Milgram's (1963) experiment, but stopped reading at the point where the learner pounds on the wall and discontinues providing answers, and then asked his (Mixon's) subjects to indicate what they believed took place from this point on. By varying relevant details in the description read to different groups of subjects, Mixon was able to show that predicted defiance increased as the harmfulness of the procedure became more certain. Thus, when the script described the experimenter as a competent person and as giving the reassurance that the shocks might cause pain but not permanent tissue damage, subjects predicted that 50% of the teachers would be fully obedient. But when, in another script, the experimenter was described as "decidedly agitated" and, rather than allaying the teacher's qualms by reassuring him that the shocks do not cause permanent damage, indicates instead that he considers the learner's health to be irrelevant, not one subject predicted full obedience (Table 3, p. 47). Greenwood (1983) also makes a similar distinction between causing pain and causing harm, arguing that Milgram went "too quickly from the assumption that his subjects were convinced they were inflicting painful electric shocks, to the assumption that they were convinced they were inflicting harmful electric shocks" (p. 242).
In a conceptual analysis of the Milgram experiments, Patten (1977) also argues for a discontinuity between obedience of the laboratory subject and obedience to commit mass murder. Patten distinguishes between the kind of authority represented by Milgram's experimenter and the kind wielded by a Hitler. The former possesses what Patten calls "expert-command'' authority. That is, he commands obedience by virtue of his presumed expertise regarding learning and shock devices. The latter, more worrisome, kind of authority wields a "simplecommand" authority, i.e., whose power to command is based on legal or quasilegal considerations, not because of any special knowledge of the task at hand. According to Patten, knowing how one might respond to expert-command authority cannot inform us about the person's behavior vis-a-vis a simple command authority.
Thus, the viewpoints just presented all share the assumption, explicitly or implicitly, that, in contrast to the murderous obedience that took place during the Third Reich, obedient subjects in the Milgram experiments typically do not believe they are harming their victim. The force of this argument is diminished, however, by the findings of two experiments that used the Milgram paradigm. In one study (Ring, Wallston, & Corey, 1970), using "painful" sound rather than shocks, 91% of the subjects were obedient-the highest rate reported in the literature for a standard obedience experiment. This high rate of obedience occurred despite the fact that the experimenter's behavior conveyed the impression of possible harm to the learner. He acted as if he had not expected the learner's complaints: At various points in the procedure, he shook his head in perplexity, and wiped his forehead.
In Munich, Germany, Mantell (1971) conducted a study, referred to earlier, consisting of a set of obedience experiments, one of which, comparable to Milgram's voice-feedback condition, yielded an 85% obedience rate. Altogether, 101 subjects served as participants. A postexperimental questionnaire revealed that 26 subjects believed that the learner was "either dead or might have died," 23 believed that he lost consciousness, and 5 believed that he suffered some sort of damage. On the other hand, 36 believed that he had undergone severe pain but "was all right," whereas 9 felt that the learner was "just fine." Thus, approximately half of the participants believed that they had harmed the victim, some with lethal consequences. The findings are clouded somewhat by the fact that Mantell does not break them down by condition nor by whether or not the subject was obedient, though he does state that "the subjects' estimates of the consequences for the [learner] correlate relatively well with the number of shocks given and therefore the number of screams heard: the more shocks given, the more severe the estimates" (p. 109).
In sum, whether one considers obedient laboratory subjects as having acted destructively, with its attendant implications about a pervasive human propensity to obey malevolent authority, depends on which arguments and evidence one gives more credence to-those consistent with Milgram's position or the ones in line with Mixon's view.

B . CONFORMITY
Milgram also made a number of important contributions to another form of social influence besides obedience, i.e., conformity. Although his conformity research broke new ground, it has rarely been recognized as such, having been overshadowed by the obedience research. Milgram's conformity experiments were groundbreaking in the following three ways. First, his cross-cultural study of conformity (Milgram, 1960(Milgram, , 1961, described earlier, was the first of its kind. Second, using an adaptation of his laboratory paradigm, he showed that peer pressure can induce an individual to give increasingly more severe shocks to a learner (Milgram, 1964a), and thereby extended Asch's line-judgment findings to a more consequential type of conformity. And, third, in a study aptly titled "Liberating Effects of Group Pressure," Milgram (1965a) showed that group pressure could yield laudable outcomes, and thereby departed radically from previous laboratory studies of conformity, which typically demonstrated the negative, constricting effects of group influence. In this variant of the obedience experiment, there were three teachers-the real subject and two confederates. Partway into the shock series, the latter defy the experimenter and refuse to continue. Fully 90% of the subjects followed suit, dropping out at some point before the end. According to Milgram (1974, p. 118), no other variant was as effective in weakening the authority's power as this one, and he draws an important lesson from his findings: "The mutual support provided by men for each other is the strongest bulwark we have against the excesses of authority" (Milgram, 1974, p. 121).
In a sense, then, Milgram provided a powerful antidote to enable individuals to withstand an authority's unwanted influence-an optimistic finding that sometimes gets lost amidst the typically gloomy textbook presentations of the obedience studies. At the same time, this study's implications were not lost on several researchers, who produced a creative diversity of studies inspired by it. Thus, Kudirka (1965), a student of Milgram, found that the experimenter's power to induce subjects to eat bitter, quinine-soaked, crackers was drastically weakened when they observed a peer (a confederate) dissenting. When subjects had to confront the experimental authority by themselves, only 26% were defiant, but when they saw a confederate drop out of the experiment, more than twice as many (58%) were able to defy the experimenter's orders. Feldman and Scheibe (1972) also showed that defiant peers could help free the subject from an experimenter's demands to carry out another kind of distasteful request-to complete a highly intrusive personality questionnaire. In one condition, naive subjects start filling out the questionnaire together with a confederate posing as a subject. Soon the confederate, visibly perturbed by the questions, announces, "This is no one's business but my own," tears up the questionnaire, and leaves the room. In a second condition, he is joined by another confederate 2 minutes later, while in a third condition, a total of three confederate-subjects, one after another, depart, without completing the questionnaire. In a fourth condition, no confederates were present in the room, and all the subjects ended up completing the experiment. But in the conditions containing confederates, as the number of dissenting confederates increased, so did the number of subjects who defied the experimenter. A more recent study (Goldstein,198 l), a doctoral dissertation supervised by Milgram, showed that peers could effectively weaken or strengthen the power of the "authority" of television. Subjects watched a series of television messages dealing with a variety of topics, e.g., film reviews and health messages, which had been taped off the air, together with a confederate. The confederate, who was either a friend or a stranger, was instructed to either agree with, disagree with, or not to comment on designated messages right before the subject was to give her judgment. Goldstein found that the peer's comments significantly modified the subject's reactions to the television messages. In the "disagreement" condition, the subject showed significantly less agreement with the TV messages than in the "no-comment'' condition, whereas the confederate's agreeing comments elicited significantly more agreement than the latter. Furthermore, friends were more effective than strangers in influencing subjects' judgments. A final instance of a study that built on Milgram's (1965a) conformity-to-defiant-peers findings was a study by Hunt (1 979) involving hypnosis. Hunt found that when a subject underwent a hypnotic induction together with a confederate who refuses to continue and leaves the room about halfway into the induction, this significantly lowered the subject's degree of obedience to the hypnotist's authority, i.e., reduced the "depth" of hypnosis he or she attained. Thus, in various ways, these four studies reflect the impact of one aspect of Milgram's work on conformity. Each study not only extended the domain of applicability of Milgram's (1965a) findings but also contains the seeds for further extension and inquiry. Thus, for example, Goldstein's (1981) television paradigm could readily be adapted to and tested for its effectiveness as a technique for counteracting antisocial and other negative forms of television fare.
A further contribution to our understanding of social influence comes from Milgram's (1974, pp. 113-1 15) conceptual analysis of conformity and obedience, yielding some clarifying distinctions between the two of them. First, obedience always involves social inequality, with the person of higher status prescribing the behavior of a person of lower status. Conformity, on the other hand, represents a person yielding to a group of peers, who have no special prerogatives to regulate his behavior. Second, conformity to a group involves imitation, whereas yielding to an authority does not. Third, although in conformity the pressure exerted is only implicit, obedience represents a reaction to explicit commands. Fourth, when people yield to group pressure, they typically deny a causal role to members of the group. In contrast, when a subject obeys the orders of an authority, the former readily admits the latter's influence.

C. CITY LIFE
Milgram's move to the City University of New York in 1967 soon produced his article, "The Experience of Living in Cities" (Milgram, 1970a), which was "seminal to the growing field of urban psychology" (Sabini, 1986(Sabini, , p. 1378. In it, Milgram applied the concept of overload to help understand adaptation to city life, and through illustrative field experiments on urban-nonurban differences in helping behavior conducted with his students, he demonstrated that it was possible to apply a disciplined, social psychological approach to the study of citytown differences. Based on an invited address Milgram gave at the annual convention of the American Psychological Association in Washington on September 2, 1969, the article achieved the status of a citation classic within a decade (Milgram, 1981b). It has been one of the most oft-cited articles in both the urban social behavior and helping literatures (Korte, 1990), and has also been discussed in some textbooks on urban sociology (Fischer, 1984;Karp, Stone, & Yoels, 1977). Building on earlier writings by the urban sociologists, Simmel (190311950) and Wirth (1938), Milgram presented the concept of overload to describe the state of the individual typically confronted by the massive, densely packed, and heterogeneous population of a city. According to Milgram (197Oa, p. 1462), the concept of overload "drawn from systems analysis, refers to a system's inability to process inputs from the environment because there are too many inputs for the system to cope with, or because successive inputs come so fast that input A cannot be processed when input B is presented. When overload is present, adaptations occur. The system must set priorities and make choices. . . . City life, as we experience it, constitutes a continuous set of encounters with overload, and of resultant adaptations." The phenomenon of the "familiar stranger" (Milgram, 1977b)-people we see regularly but never interact with, such as fellow commuters at a train station-represents one such adaptation.
Why has the concept caught on? A possible reason, which I believe has a lot of merit, has been articulated by Little (1987, p. 222): There is a general thesis of considerable popularity within environmental psychology to the effect that humans have limited capacity for processing information, that contemporary environments, particularly urban ones, provide a surfeit of such information, and that this situation in part creates much of the malaise of current living. Certainly the most influential statement of this thesis was Milgram's (1970a)  "The Experience of Living in Cities" has been influential both in theory development and in stimulating research. The overload concept is one of the underpinnings of Altman's (1975) theory of crowding. According to Altman, crowding takes place when various privacy-regulation mechanisms fail to attain our desired level of privacy. Geller (1980), a student of Milgram, added the concept of optimal level of stimulation to the overload concept to produce a more "balanced" formulation, which sees urban stimulation as potentially resulting in positive as well as negative outcomes. And Cohen's (1978) approach to the study of urban and environmental stressors extended Milgram's model to encompass physical as well as social sources of overload. In a recent assessment of the overload concept, Korte (1990) states that the 1970s marked a significant increase in research conducted on urban-nonurban differences in social behavior and expresses the belief that many of the researchers involved had been inspired by Milgram's work, often as a result of studying with him as his graduate students.
What does the future hold for Milgram's overload concept? According to Korte (1990, p. 8), it "may depend on the occurrence of additional insights and understandings of this fascinating notion. Since computer systems were in part an inspiration to the thinking of Milgram and others on this point, perhaps future developments in computer systems will revitalize and reshape the overload concept. The field of urban social psychology has been enriched by Stanley Milgram and the input overload concept, as well as by the research that has resulted in this area, and I am hopeful that this contribution will prove to be of lasting value." Milgram's "cognitive maps" studies reflected another aspect of his interest in the psychology of urban life. In fact, Milgram's (1970a) article ends with a section titled "Cognitive Maps of Cities," which essentially previews the technique used in Milgram et al. (1972) in which New Yorkers were asked to identify photographs of geographically sampled locations throughout the five boroughs of New York City. Although not nearly as influential as his urban overload concept, "A Psychological Map of New York City"  brought increased precision and rigor to the study of the "imagability" of cities pioneered by the urban planner, Lynch (1960); its findings, together with those of the methodologically richer cognitive-mapping study of Paris (Milgram & Jodelet , 1976, represent substantive contributions to the accumulated body of knowledge on environmental cognition (for reviews, see G. W. Evans, 1980;Golledge, 1987).

D. TELEVISION AND ANTISOCIAL BEHAVIOR
When Milgram turned to the study of television violence (Milgram & Shotland, 1973), he did so in a manner that remains unique to the present day. Virtually all experimental research on the effects of televised antisocial behavior uses already-existing fare as stimulus materials. In a typical study, subjects are randomly assigned to watch either a program that is clearly violent in content or one that is free of violence, and are subsequently compared on some dependent measure of aggression. A built-in problem with this approach is that the chosen programs may also differ in other qualities besides the presence or absence of antisocial themes. Milgram and Shotland (1973) departed radically from this traditional approach by manipulating the independent variable via the systematic variation of the content of one episode of a popular prime-time television series. Specifically, Milgram was able to get the cooperation of CBS to produce a particular episode of the dramatic series, Medical Center, with three different endings. In that episode, the plot revolves around the misfortunes of a hospital orderly, Tom Desmond: He loses his job, with a sick wife and an infant to support, and his boat, which he had hoped would provide some income, is repossessed. %o versions end with a destructive act, the orderly repeatedly smashing open fund-raising collection boxes distributed by the Medical Center and stealing the money they contained. In one variant he is caught by the police; in the other, he eludes capture. The third version had a prosocial ending, with Desmond putting money into, rather than stealing from, a collection box. The two antisocial versions also showed the orderly placing two abusive phone calls to the medical director of the hospital. Another episode, with a completely different storyline, without any suggestion of violence whatsoever, served as a baseline, control condition. Milgram and Shotland embedded these programs in a series of field experiments in which both viewing and the opportunity to imitate the antisocial acts occurred in real-life setting^.^ The study is not without shortcomings. One of the eight experiments, for T h e first seven experiments used the Medical Cenrer episodes. In order to determine whether depicting the antisocial act as an actual, rather than a fictional, event made a difference, the eighth study embedded it in a videotape of a local newscast. example, recruited subjects from New York's Times Square area, a group that apparently contained vagrants, alcoholics, and drug addicts-not a very generalizable sample. I have a copy of the prosocial version of the Medical Center episode used by Milgram and Shotland. I regularly show it to my social psychology classes and also in a winter session course on the social psychology of Stanley Milgram which I have created (Blass, 1991~). Toward the end of the film, whenever Desmond fixes his gaze on a charity box, his potential prey, the soundtrack accompanies it with a "boing-boing" sound to bring it into dramatic focus. But the sound, not unlike the kind one hears when a cartoon character bounces back after running into a brick wall, creates an unintended melodramatic effect, as evidenced by the laughter I sometimes hear from my students as they watch that scene. Also, as a number of writers have correctly pointed out, no theoretical context is provided by Milgram and Shotland for their study and only minimal attention is given to prior relevant research on observational learning and media influence (see Comstock, 1974;Dominick, 1975;Feshbach, 1975). Nevertheless, its unique methodology and scope (a total of eight integrated field experiments and, in one of them, millions of at-home viewers as potential subjects) make its null-effects findings deserving of special attention in any discussions of media influences. Yet a look at reviews of the literature on TV violence and viewer aggression shows that Milgram and Shotland's research program has had very little impact on the debate about the harmful effects of television. Most reviews, if they mentioned it at all, have dismissed it because of methodological problems (Comstock, Chaffee, Katzman, McCombs, & Roberts, 1978;Geen & Thomas, 1986;Huesmann, 1982;Liebert, Sprafkin, & Davidson, 1982; but see also Freedman, 1988;Kaplan & Singer, 1976, for more favorable reactions to the study). In focusing largely on the flaws in the Milgram-Shotland television experiments, these reviews overlooked what their unique design was able to achieve-a degree of control over the independent variable rarely attained in experiments on the effects of TV violence, a design feature that is crucial if one wants to be able to draw causal conclusions.

E. THE LOST-LETTER TECHNIQUE
In a brief note, Milgram (Milgram et a]., 1965) introduced the lost-letter technique as a method "for assessing community orientations toward political groups and other institutions" ( p. 437). The technique involves "losing" stamped, addressed envelopes in various locations-on sidewalks, in stores, telephone booths, and under windshield wipers of cars. By varying the name of the addressee on the letters, and comparing the proportions of the letters mailed to each target, prevailing attitudes toward various groups or organizations can be measured unobtrusively. The technique had originated with Merritt and Fowler (1948); however, they used it to study the personal honesty of the finders of the letters, whereas Milgram's innovation was to use rate of returns "as a clue to how people felt and-more important-how they would act toward different political organizations" (Milgram, 1969c, p. 3 1).
.In the first experiment, conducted in New Haven, Connecticut, 400 letters were distributed, with 100 letters each bearing the following designations: Friends of the Communist Party, Friends of the Nazi Party, Medical Research Associates, and a Mr. Walter Carnap. The number of letters returned was 25, 25, 72, and 71, respectively. Milgram and his students went on to use the technique to predict the winner of the 1964 Presidential election, to study attitudes toward racial integration in Connecticut and North Carolina, to determine the political allegiances of the Chinese residents of Honk Kong, Bangkok, and Singapore, and to identify pockets of neo-Nazi sentiment in Munich, Germany .
The lost-letter technique has been the most widely used nonreactive measure of attitude and opinion. By 1981, over 30 studies had used it (Webb, Campbell, Schwartz, Sechrest, & Grove, 1981; see it, as well as Sechrest & Belew, 1983, for reviews), and it continues to be used for a variety of purposes, from studying the Catholic-Protestant conflict in Northern Ireland (Kremer, Barry, & McNally, 1986) to measuring attitudes toward abortion (Kunz & Fernquist, 1989). Perhaps more important than the number of studies that have employed it is the fact that, more generally, the lost-letter technique has helped sensitize social psychologists about the value of nonreactive measures in attitude measurement.
The technique has a number of limitations, some of which were identified by Milgram (1969aMilgram ( , 1969c: There is no control over the mediating processes leading to the return of the letters; it should work for highly polarized and emotionally charged issues but not for subtle ones; because a lot of uncontrolled yariance is involved, response rates cannot provide exact information about population parameters, and for the same reason, large numbers of letters need to be distributed in order to be able to obtain between-condition differences; its primary usefulness is when the respondents' awareness that they are being studied would seriously distort their answers, an opinion echoed in Sechrest and Belew's (1983) cautionary evaluation of the lost-letter technique: It seems like a relatively simple nonreactive measure on the surface, but factors such as location of the drop sites, preservation of a stamp, potential curiosity arousal of the finders in response to some addresses, and the possibility of money inside the envelopes all can affect returns and interfere with concluding anything but finders' attitudes. However, the method should not be disregarded completely. Ir very well may prove to be rhe optimal method, or at least a useful supplement to other methods. when rhe issues for which attirudes are being assessed are of a highly sensitive and rhus reacrive narure. What is absolutely essential, however, is that potential users of the technique do systematic methodological testing before employing the measure for actual attitude assessment. (p. 42; emphasis added)

F. THE SMALLWORLD METHOD
In 1979, communication researcher Everett M. Rogers was a visiting professor at a university in Mexico City. One day, while having a conversation with a student named Pedro Felix Hernandez, Rogers complimented him on the excellence of his English. The student replied that this was a result of having lived with a family in Iowa when he was there as an exchange student. The conversation continued (Rogers & Kincaid, 1981, p.  gation of the diffusion of 2.4-D weed spray! Two people meeting for the first time and discovering that they have someone in common-as illustrated in the just-quoted conversation between the professor and the student-happens surprisingly often, and has been referred to as the small-world phenomenon (Kochen, 1989). The more general question these kinds of encounters pose is the small-world problem, which Milgram (1967b) formulated as follows: "Starting with any two people in the world, what is the probability that they will know each other? . . . Another question one may ask is: Given any two people in the world, person X and person Z, how many intermediate acquaintance links are needed before X and Z are connected?" ( p.

62). lo
To provide an answer, Milgram devised the small-world method, which involves the following features: The name of a randomly selected person (the target), usually in a distant city, is given to a sample of men and women (starters). The task of the starters is to send a folder to the target person using only a series of friends and acquaintances who would be more likely to know the target person than they (the starters) would. The folder could only be sent to a person 'There is a slightly longer version of this article (Milgram 1969b). with a different set of illustrations, including a reproduction of the contents of the folder that was used in the mailings. whom the sender knows on a first-name basis. To keep track of the course of the folder, it contained a roster to which each subject had to add his or her name and tracer postcards to be mailed to Milgram. Among the main findings obtained by Milgram and colleagues (Korte & Milgram, 1970;Milgram, 1967b;Travers & Milgram, 1969) using the smallworld method, were the following: 1. The average number of intermediaries in completed chains ranged from 4.4 to 5.9. 2. Subjects were much more likely to send the mailing on to friends or acquaintances than to relatives. 3. Only a fraction of the chains were completed. For example, in a study with starters in Nebraska and a stockbroker in Boston as the target person, only 24% of the mailings eventually reached him. 4. In an experiment composed of white starters in Los Angeles who had to reach either a white or black target in New York City, a chain was two and a half times as likely to reach the former than the latter.
Milgram's reports on the small-world method stimulated a good deal of research, and, by 1979, there were already enough studies using the technique to merit a review article (Bernard & Killworth, 1979). Although a few social psychologists other than Milgram and his colleagues have used the small-world method (e.g., Bochner, Duncan, Kennedy, & Orr, 1976;Shotland, 1976), its primary impact has been among social-network researchers, a group composed largely of sociologists, political scientists, and anthropologists. Killworth and Bernard (19781 1979), for example, were inspired by the small-world method to create their own technique for examining social networks, called the reverse small-world technique. In this method, respondents are given a large number of targets and are asked to indicate their choice of the first link in a potential chain to reach each target. In another modification of the small-world method, Guiot (1975) substituted telephone calls for mailings, and achieved a much higher completion rate (85%) than did Milgram and colleagues.
Studies of social networks have typically relied on self-report measures, e.g., questions asking subjects to list individuals with whom they have a particular kind of relationship (Marsden, 1990). Through the small-world method, Milgram has enabled network researchers to use subjects' actions as a source of information about their social ties.
Perhaps the most salient indicator of the degree of influence of the small-world method on the field of social-network research is the recent publication of an edited volume, The Small World (Kochen, 1989), which has as its subtitle, "A volume of recent research advances commemorating Ithiel de Sola Pool, Stanley Milgram, Theodore Newcomb." (Unpublished working papers by Pool and Kochen presenting a theoretical model of the small world had stimulated Milgram's interest in the problem.) A number of the chapters in the book deal with the small-world method. Kochen (1989) describes Milgram's (1967b) first article on the subject as a "seminal paper" (p. ix), and states that the technique "established an experimental paradigm for this field" (p. 149, and Kadushin (1989) notes that it "remains one of the critical tools of network analysis" (p. xxiv). A chapter by Klovdahl(1989) explores the limitations of the small-world method (see also B. H. Erickson, 1978).

V. Conclusions
American psychology has had a love-hate relationship with Stanley Milgram. Although his obedience research is a classic of modem psychology and a "must" topic for the introductory-level course, the American Psychological Association never gave him its Award for Distinguished Scientific Contributions, an award it has given to several important social psychologists over the years. I I Social psychologists will invariably invoke the results of the obedience experiments whenever we need to affirm that our field can reveal something about social behavior that is not predictable from common sense. Yet most writings on the history or theories of social psychology make only passing mention of Milgram-if they mention him at all. These lacunae would seem to be at odds with the fact that references to his writings by scholars and researchers continue unabated: An on-line bibliographic search using the Institute for Scientific Information's "Social Science Citation Index" data base reveals that, through October, 1991, he was cited 46 times in 1991 alone and a total of 1720 times since 1972.
Yet, Milgram must take at least part of the blame if the conventional, "establishment" forms of recognition eluded him. I believe it had a lot to do with the impression he created among some psychologists of a dilettante who flitted from one newsworthy phenomenon to the next, not staying with any one long enough to probe it with adequate depth. Even those with a more favorable view of Milgram's accomplishments are not likely to think of him foremost as someone who did programmatic, time-consuming research. Yet he did exactly that, three different times in his career. His obedience research comprised over 20 interrelated studies spanning the years 1960 to 1963. The television study, consisting of eight sequential field experiments, was first proposed to CBS on April 23, 1969. Most of 1970 was devoted to correspondence with the writer of the Medical "He did win, in 1964, the annual Socio-Psychological Prize of the American Association for the Advancement of Science for his article "Some conditions of obedience and disobedience to authority" (Milgram,196.5~).
Center episode that was to serve as the stimulus program and to the actual production of the film, with Milgram sitting "on the sidelines at MGM studios acting as a censor whenever an element was introduced that did not meet the needs of the experiment" (Milgram & Shotland, 1973, p. 74). The experiments themselves were conducted from September, 1970, through November, 1971. To conduct his dissertation research on conformity, Milgram spent 1957-1958in Norway and 1958-1959 in France, carrying out a total of 14 experimental variations. Including subjects who participated in pretesting the procedures at Harvard during the summer of 1957, a total of 390 participants served in the conformity experiments (Milgram, 1960).
Milgram was equally at home publishing in magazines as in journals, which made him an effective disseminator of psychological information to the public. At the same time one sometimes almost got the feeling that for Milgram magazines such as Psychology Today (e.g., Milgram, 1977d) and Omni (Milgram, 1984c) were functionally equivalent to the Journal of Personality and Social Psychology. Thus, the first article about the small-world method was published not in a scientific journal but in Psychology Today (Milgram, 1967b). I believe that similarly odd publication decisions were at least partially responsible for two of the three aforementioned research endeavors not getting their proper recognition as sequentially developed programmatic efforts. I have already mentioned, in a previous section, the piecemeal nature of the publication of the obedience research: It took Milgram a full 9 years after the appearance of the last journal report on the obedience work  to reveal (in Milgram, 1974) the complete research program, which included experimental conditions not described in earlier publications. With his cross-cultural research on conformity (his doctoral dissertation, Milgram, 1960), Milgram completely flouted the norms of scientific publication because the only place he published it was in Scientific American (Milgram, 1961). Because readers do not normally expect a magazine to be the primary publication vehicle for the presentation of original research findings, for a long time many psychologists remained unaware of the conformity studies and the extensive program of research they represented.
Despite the lack of attention to Milgram in most writings on the history of social psychology, he can readily be placed in a historical context. His phenomenon-centered approach represents a continuation, through Asch, of the Gestalt tradition. At the same time, his boundless confidence in the possibility of studying a wide range of social phenomena scientifically makes him supremely Lxwinian.
The innovative work of some of Milgram's students and colleagues bears his imprint. Examples that come to mind are Gaertner and Bickman's (1971) wrongnumber technique; Korte's operationalization of the urban overload concept (Korte, Ypma, & Toppen, 1975); Takooshian, Haber, and Lucido's (1977) lostchild study; and Mann's (1981) attempt to account for the formation of baiting crowds at episodes of threatened suicide. The work of Sabini and Silver (e.g., Sabini, 1976;Sabini & Silver, 1982Silver, 1977) has continued and amplified Milgram's focus on the moral aspects of social behavior. Milgram also had a lasting influence on his students in less easily identifiable ways. For example, Judith Waters, a former student and research assistant, told me that every time she conducts a study she asks herself: "Is this how Stanley would have done it?" Milgram spent most of his professional career (1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983)(1984) at the Graduate Center of the City University of New York. During that time, he supervised more doctoral dissertations in social psychology than any other CUNY faculty member. Many more students undoubtedly came into contact with him through coursework, seminars, or as a member of their dissertation committees. Yet his students have not had the kind of impact on social psychology as have those of Festinger and Schachter. 1 am not sure why this is, but one possibility is contained in a tribute by one of Milgram's former students, which she wrote in the acknowledgments section of her doctoral dissertation: Although the late Stanley Milgram did not live to see this dissertation completed. he brought an element of realism and vision to this work. Stanley understood the strengths and limitations of social psychology. Likewise, he had a true appreciation of the strengths and limitations of human beings, and had the wonderful ability to elicit the most out of both his students and science. Although he had dijjiculfy in transmitting his innate creativity to students, as a teacher-exemplar he taught me basic tools of social psychology. But most of all I will cherish Stanley's altruistic nature and the positive reinforcement I received from him. (Fogelman, 1987, p

. ix, emphasis added)
Milgram sensitized us to the hidden workings of the social world. He showed the difficulty people often have of bridging the gap between intentions and actions. Even moral principles are not invariably translated into behavior but can have their potential power overridden by momentary situational pressures. Although people can be responsive to the precise and subtle details of the immediate situation, people are generally not aware of their power and typically think of themselves as free of social pressures. Social norms can often have a compelling effect on our behavior, wielding their power by means of the unexpected amount of inhibitory anxiety generated by their violation.
While some critics were chipping away at the bedrock findings of social psychology through claims of bias and artifact, Milgram was expanding its proper domain by turning his attention to such topics as photography, Candid Camera, mental maps of cities, and subway norms. To dispel the gloom created by periodic pronouncements about a crisis in social psychology, one could always share vicariously in the "pleasurable activity of experimental invention" (Milgram, 1977e, p. 2) through exposure to such innovations as the lost-letter technique and the small-world method.
What is remarkable is that despite the pessimism occasioned by the obedience findings, Milgram was able to maintain a positive, hopeful view of human potential: "Though enormous pressures may bear down on a person to abandon critical intelligence, dispense with conscience, and surrender humanity, that person will often prove hardy and resilient, transcend the pressure of the moment, and reaffirm the power and integrity of one's own spirit. Well, as our experiments show, it does not always happen this way. But it remains a worthy ideal" (Milgram, 1977e, p. 155). This, perhaps, is his greatest legacy.