Another Look at the Milgram
Steven J. Gilbert
Milgram's (1963, 1965a, 1965b) incremental shock procedure for quantifying
obedience may be partly responsible for the high levels of obedience
7he innocuous beginning of the shock sequence Dow voltage,
no negative feedback) may elicit compliance before the frightening
implications of the procedure are clear, and the gradual escalation
in shock intensity may deprive subjects of a qualitative breakpoint
needed to justify changing from obedience to disobedience.Parallels between
the incremental shock and the 'foot-in-the-door" (Freedman & Fraser,
1966) techniques for obtaining compliance are discussed.
Milgram's (1963) original obedience study is, perhaps, the best known product of experimental social psychology. Who among us was not surprised and sobered to learn that 65% of his subjects delivered the full series of painful and escalating shocks to an innocent partner? Milgram painted an unforgettable picture of a psychic battle raging between the subject's compulsion to obey the directives of a legitimate authority, and his impulse to stop hurting the victim he has been ordered to punish.
Milgram's study was actually one in a series of experiments (Milgram, 1963, 1965a, 1965b, 1974) designed to "examine the situational variables responsible for the elicitation of obedience" (Milgram, 1963, p. 73). A constant feature of these studies was the use of a shock generator containing 30 switches, each ostensibly delivering 15 more volts of shock than the one before. All subjects were instructed to deliver the next in the gradated series of shocks whenever the learner made an error . This arrangement was chosen in part because it allowed a quantitative measure of the dependent variables 30-point response scale in which "maximum level of shock delivered" was taken as a measure of the construct "level of obedience."
The gradated shock board also facilitated the fashioning of independent variable manipulations. By requiring the subject to deliver increasing levels of shock, Milgram was able to credibly attach increasingly negative and extreme reactions from the victim to successive shock administrations by the subject. Had all the shocks been ostensibly equal in intensity (and presumed painfulness), such a manipulation would have been far less convincing.
In his original pilot work, Milgram's subjects received no feedback from the victim. This situation produced virtually perfect obedience (Milgram, 1965b). The surprising results led Milgram to devise a number of experimental conditions in which he varied the level of "immediacy" of the victim-the extent to which the pain, potential injury, and unwillingness of the victim to continue receiving shock were made obvious to the subject. Milgram hypothesized that the sooner a subject received such negative feedback from the victim (that is, the earlier in the shock series), and the "psychologically closer" that the feedback brought the subject to the victim's pain, the less likely the subject would be to show complete obedience, and the sooner he would show disobedience. The results of these studies clearly suppor ted this hypothesis (Milgram, 1965b).
It is important to note, however, that even in the condition involving
the greatest immediacy (Touch
The results suggest that while the immediacy experiments shed light on factors which quicken disobedience in Milgram's experimental situation, they do not explain the power of the situation itself to produce obedience, and to maintain it in the face of powerful contravening forces. It can be hypothesized that this power largely resides in one of the features of Milgram's paradigm which he did not vary-the gradated shock procedure which stands at the heart of all of his obedience experiments.
In light of contemporary ethical standards, an experimental demonstration of the role of the gradated shock procedure in producing obedience is unlikely, and probably undesirable. However, other lines of evidence can be pursued. For example, the unique capacity of this procedure to elicit obedience can be highlighted in a hyp othetical experiment. Imagine an alternative design, in which there are no gradations of shock (or obedience). After receiving considerable word association training from the subject, the learner is given a major test, and fails. The subject must now deliver one "extremely dangerous" shock of 450 volts, 10 times the voltage of the sample shock which all subjects in these experiments receive. Do 65% of these subjects deliver the 450 volt shock, as in the 1963 experiment? Or 100%, as in the earlier pil ot studies?
My speculation is that fewer subjects would obey, because obedience would entail a single, clear, and strongly contra-attitudinal act. Delivering this shock would be inconsistent with and qualitatively different from anything the subject is likely to have done before, and recognized as such by the subject. Indeed, anticipating that the learner might fail, and that this exigency would require the administration of a single, horribly painful, and possibly dangerous shock, it is pr obable that many subjects would refuse to begin the teaching session at all.
The gradated shock method regularly used by Milgram avoids a single, explicit confrontation of values. Instead, the subject is carefully shaped into obedience. Indeed, the first step-delivering a small shock after a foolish mistake by the learner-may not feel particularly repugnant to subjects, because the implications of the act are unclear to them. As they press the first switch, they may not anticipate that the learner will do badly enough to require many more shocks, or that one shock will seem to "occasion" the next, or that delivering a few shocks will feel so much like publicly declaring acquiescence to the requirements of the experiment, or that such an apparent declaration will feel so personally binding (though made in functional ignorance of its consequences).
Milgram aptly characterized his obedience method as one which causes the subject to become "integrated into a situation that carries its own mom entum. The subject's problem ... is how to become disengaged from a situation which is moving in an altogether ugly direction" (1965b, p. 73). In essence, I have suggested that the gradated shock procedure may have two important consequences: (a) it engages subjects in committing precedent-setting acts of obedience before they realize the "momentum" which the situation is capable of creating, and the "ugly direction" in which that momentum is driving them; and (b) it erects and reinforces the impressi on that quitting at any particular level of shock is unjustified (since consecutive shock levels differ only slightly and quantitatively). Both consequences of the gradated shock procedure may conspire to deprive subjects of the credible rationale they need to quit at any given point before completing the experiment.
If the gradated shock procedure is crucial in eliciting and maintaining obedience in the Milgram steadies, then it follows that when disobedience does occur, it should happen most oft en when the learner has just received feedback from the victim suggesting that there has been some qualitative change in the latter's experience of the shocks. it is when the subject believes that the next shock will really hurt (rather than simply be another degree more unpleasant), or that the learner now wants to get out (rather than seeming simply a JND sorrier that he had volunteered), that subjects should refuse to deliver any more shocks.
Milgram's results appear to be consistent wit h this expectation (1963, 1965b). For example, in the Remote Feedback Condition (Milgram, 1963) subjects received no feedback from the victim until hearing pounding on the wall at 300 volts. None of the 40 subjects in this condition discontinued giving shocks prior to receiving feedback at the 300-volt level. Of the 14 subjects who did not show full obedience, 5 quit after the wall pounding episode-the first qualitative change in the apparent consequences of the behavior of the subject. Four subjects stopped delivering shocks after 315 volts, when they discovered that the victim would (or could) no longer give word association answers-the second qualitative change in the effect of shocking the learner. Thus, 9 of the 14 subjects (65%) who disobeyed at all, disobeyed in response to information suggesting that delivering an additional increment of shock would have qualitatively different consequences for the victim, and a qualitatively different meaning for the perpetrator.
Another line of evid ence supports the contention that much of the obedience eliciting power of Milgram's experimental situation inheres in the capacity of the gradated shock procedure to channel subjects' perceptions of their actions into a unidimensional series containing no decisive, quantitative break-points. Those familiar with Milgram's original report (1963) will recall that his subjects often questioned the experimenter and asked to be allowed to stop giving shocks. The procedure called for the experimenter to re ply in a calm, but quietly assertive manner, with a series of "prods"-phrases such as "please continue," and "the experiment requires that you continue."
Mixon ( 1 972) points out that the tone and content of these prods imply that the victim is not being hurt more than participants in the experiment should expect or allow. This apparent unconcern of the experimenter may serve to disqualify or supersede the warnings which appear on the shock board (such as "Danger: Severe Shock"), 'and the feedback coming from the victim (such as screaming, wall pounding).
Mixon's speculations appear to be consistent with the results of a roleplaying simulation of the Milgram technique which he performed (I 972). Given a description closely resembling Milgram's 1963 procedure (DM-4), 50% of the participants predicted perfect obedience. In a second condition (DM-6), subjects received an identical description of Milgram's procedure, except that instead of maintaining a calm and deliberate manner, the experime nter was described as responding to the escalating levels of negative feedback emanating from the victim with signs of increasing surprise, agitation, and worry. Only 20% of the participants in this condition predicted perfect obedience.
Some of the capacity of Mixon's DM-6 condition to "release" disobedience may have emanated from the redefinition it provided of the meaning of the act of delivering the next shock. The surprise and worry shown by the experimenter may have legitimized the interpre tation that a qualitative change has taken place in the meaning of the act of pressing the shock switches. The next shock now could be judged as involving more than a simple continuation of what the subject seems to have already committed himself to. Instead, it now may be viewed as sufficiently different to justify breaking the implicit agreement with the experimenter, that "I'll keep doing this (delivering escalating shocks) as long as you keep proving that I'm not being made to do something new, d ifferent, or really worse."
Milgram's gradated shock method may share important features with other instances o' successful persuasion and behavior change. These include systematic desensitization using an anxiety hierarchy (Wolpe & Lazarus, 1968), the "utilization" technique in hypnosis (Haley, 1963), and selling techniques involving eliciting escalating commitments to buy (Varela, 197 1). On a broader scale, mechanisms akin to Milgram's use of gradated shock may be frequently used by psychot herapists (Goldfried & Davison, 1976, pp. 61-64), and may have been manifest in the escalation of the Vietnam War through gradual troop increases, and tragic obedience demonstrated at Jonestown (Osherow, 1981).
Within the mainstream of experimental social psychology can be found at least one other classic study which used a compliance-inducing method resembling Milgram's gradated shock technique. In the "foot-in-the-door" experiment of Freedman and Fraser (1966), women in one group were asked to allow an investigator to place a very large and ugly sign on their front lawns for a week, and 25% complied. Women in a second group were first approached with the request that they answer a few innocuous questions (virtually all complied) and then were given the sign-on-lawn request three days later. As hypothesized, compliance was significantly greater for the second group-55% of these women agreed to have the sign placed on their lawns.
The single-request condition in the Freedman and Frase r study resembles the hypothetical alternative to the Milgram gradated shock method described previously. In both "one shot" methods, compliance with an experimenter's request involves a single act with obvious consequences which clearly violate values and sensibilities (ugly sign; harmful shock), and which is probably qualitatively different from anything done by the subject in the past.
The second condition of the Freedman and Fraser study resembles the gradated shock procedure used by Milgram i n the obedience research. Once again, the first step toward compliance (answering a few questions) was easily obtained because the implications of the act were unclear-in fact, unknowable. Despite the large quantitative differences in the scope and costliness of the first request (answering questions) and the second (allow ugly sign to be put on lawn), they may have been perceived by the subject as two instances of the same thing. Both involve having vs. having to refuse to help a stranger who appear s to be acting under credible motives. Having psychologically declared themselves "helpers of strangers asking for a legitimate favor" by complying with the first request, many subjects may not have been able to quickly and confidently find a qualitative difference between the first and second request sufficient to justify refusing the latter. Indeed, these women may have found themselves involved in "a situation that carries its own momentum." Unable to define away the essential similarity of the second request to the first, they failed to "disengage from a situation [and a sign] which was moving in an altogether ugly direction."
The parallels between the Milgram (1963) and Freedman and Fraser (1966)
experiments are not discussed in 23 social psychology texts I have sampled
(e.g., Aronson, 1980; Gergen & Gergen, 198 1; Freedman, Sears, &
Carismith, 198 1; Shaver, 1981; Seidenberg & Snadowsky, 1978). Typically,
the Freedman and Fraser experiment is explained in terms of the
of its unique two-stage compliance method (foot-in-the-door technique)
upon subjects' coding of the situation, their self-perceptions, and their
resultant behavior. In contrast, discussions of the Milgram experiment
ignore the role of its gradated compliance method (the finger-on-the-switch
technique?) in channeling situational definitions and self-perceptions
into behavioral obedience. It is hoped that textbook authors will come
to recognize that explication of the role of the gradated shock
is necessary for a full appreciation of Milgram's disquieting revelations
of how much and how easily obedience to authority can be obtained.
Aronson, E. The social animal San Francisco: W. H. Freeman, 1980.
Freedman, J. L., & Fraser, C. C. Compliance without pressure: The foot-in-the-door technique. Journal of Personality and Social Psychology 1966, 4, 195-202. Freedman, J. L., Sears, D. O., & Carlsmit h, J. M. Social Psychology. Englewood Cliffs, NJ: Prentice-Hall, 1981.
Gergen, K. J., & Gergen, M. M. Social psychology. New York: Harcourt Brace Jovanovich, 1981.
Goldfried, M. R., & Davison, G. C. Clinical behavior therapy. New York: Holt, Rinehart,and Winston, 1976.
Haley, J. Strategies of psychotherapy. New York: Grune & Stratton, 1963.
Milgram,S. Behavioral study of obedience. Journal of Abnormal and
Social Psychology, 1963, 67,
Milgram, S. Liberating effects of group pressure. Journal of Personality
and Social Psychology, 1965,
Milgram, S. Some conditions of obedience and disobedience to authority.
Human Relations, 1965, 18,
Milgram, S. Obedience to authority: An experimental view. New York: Harper & Row, 1974.
Mixon, D. Instead of deception. Journalfor the Theory of Social Behavior, 1972,2, 145-177.
Osherow, N. Making sense of the nonsensical: An analysis of Jonestown. In E. Aronson (Ed.), Readings about the social animal. San Francisco: W. H. Freeman, 1991.
Seidenberg, B., & Snadowsky, A. Social psychology. New York: The Free Press, 1976. Shaver, K. G. Principles of social psychology. Cambridge, MA: Winthrop, 1981.
Varela, J. A. Psychological solutions to social problems: An introduction
to social technology.
Wolpe, J., & Lazarus, A. A. Behavior therapy techniques. New York: Pergamon, 1968.
Steven J. Gilbert is an Assistant Professor of Psychology at the State University of New York College at Oneonta. His research interests include compensatory causal schemas, persuasive techniques of advertising, and levels of interpersonal perception and judgment.