Compliance is produced by organizing the conditions under which people judge and act. Apparent consensus makes individuals distrust their own perception. Authority defines obedience as responsible conduct. Roles narrow what each participant is expected to consider. Procedures divide harmful outcomes into small, defensible steps. Distance conceals the human consequences. Incentives reward cooperation and punish refusal.
These pressures do not erase personal agency. They shape how a decision appears, which choices seem available, and what each choice will cost. Working together, they can lead people to support outcomes they would reject if required to confront the full act and accept personal responsibility for it. Understanding compliance therefore requires examining the structure that makes obedience easier to justify and resistance harder to sustain.
Harm Divided Into Steps
In 1961, 40 men entered a Yale University laboratory believing they were participating in research on learning and memory. Each was instructed to administer an electric shock to another man whenever he answered a question incorrectly.
The shocks were simulated, but the participants believed they were real. The machine contained 30 switches beginning at 15 volts and ending at 450, with labels progressing from “Slight Shock” to “Danger: Severe Shock.” At 300 volts, the man in the adjoining room pounded on the wall. After 315 volts, he stopped responding.
Every participant continued to at least 300 volts. Twenty-six of the 40, or 65 percent, reached the maximum setting.
The men were not physically restrained or unaware of the apparent danger. Many sweated, trembled, protested, and asked to stop. Most continued when the experimenter calmly insisted that the procedure required it.
Milgram’s experiment showed how a controlled environment could make continued obedience easier to justify than refusal. Yale supplied institutional credibility. Science supplied a legitimate purpose. The experimenter appeared to possess expertise and accept responsibility. The participant performed a limited technical task while remaining physically separated from the person receiving the supposed shocks.
The voltage also increased in increments of only 15 volts. Participants never faced the entire act as one decision. Each choice concerned the next switch and followed a long series of decisions they had already accepted. Stopping required them to admit that the procedure, the authority directing it, and their own previous participation had become indefensible.
No single feature forced them to continue. Together, the conditions raised the immediate cost of refusal and repeatedly supplied reasons to proceed. Real institutions use the same pressures. Harmful conduct arrives as a policy, assignment, performance target, professional standard, security measure, or established procedure. Participants are asked to perform one authorized part of an outcome divided among many people.
Consensus Raises the Cost of Independent Judgment
Solomon Asch’s conformity experiments tested whether people would publicly contradict an obvious fact when everyone around them gave the same incorrect answer.
Participants were shown one line and asked which of three comparison lines matched its length. Under ordinary conditions, errors were rare. In the experimental setting, each real participant answered after several other group members, all of whom had been instructed to choose the same incorrect line.
Across trials involving 123 participants, subjects adopted the majority’s incorrect answer on 36.8 percent of the critical trials. Three-quarters agreed with the incorrect majority at least once. About one-quarter remained independent throughout.
The group changed the cost of trusting personal judgment. Some participants assumed that the others understood something they did not. Others knew the group was wrong but wanted to avoid embarrassment, conflict, or isolation. Public disagreement required them to place their own perception above a unanimous social signal.
The size of the majority mattered less than its apparent unity. One incorrect opponent produced little pressure. With two, the error rate rose to 13.6 percent. With three, it reached 31.8 percent, after which additional group members made relatively little difference. The majority did not need to be large. It needed to appear settled.
That appearance could be disrupted by one person. When a partner gave the correct answer, conformity fell to roughly one-quarter of its previous level. Even a partner who rejected the majority while offering a different wrong answer reduced its influence. The decisive change was the loss of unanimity.
A 2023 replication by Axel Franzen and Sebastian Mader found a 33 percent error rate when 210 participants faced five confederates in the standard line task. Paying participants for correct answers reduced the rate to 25 percent but did not eliminate conformity. In a separate task involving political statements, the conformity rate reached 38 percent.
The strength of this pressure varies across societies and historical periods. A meta-analysis of 133 Asch-type studies conducted in 17 countries found substantial cultural and historical differences. People respond to the meaning that agreement, dissent, and social exclusion carry in their environment.
When people repeat an approved position in public, observers often mistake visible conformity for genuine belief. Those who privately disagree see no sign that anyone shares their judgment, so they remain silent. Their silence strengthens the appearance of consensus and increases the pressure on everyone else.
This cycle allows authority to present public compliance as evidence of legitimacy. A dissenter interrupts it by making disagreement visible. One person cannot end an organized system through expression alone, but they can remove the false belief that everyone agrees and lower the social cost of honesty for others.
Authority Makes Obedience Appear Responsible
Milgram added legitimate authority to the pressure of the situation. His participants were adults between 20 and 50 years old, drawn from different occupations in the surrounding community. Before the experimenter issued an instruction, the Yale setting, laboratory equipment, and professional appearance of the researcher established expectations of expertise and legitimacy.
When participants hesitated, the experimenter used calm, standardized prompts. The language framed continuation as a procedural obligation rather than a new moral decision. The participant could therefore understand obedience as cooperation with a competent authority and refusal as interference with valid scientific work.
Milgram described an “agentic state” in which people experience themselves as instruments carrying out another person’s wishes. Albert Bandura’s work on moral disengagement identified related ways people separate their conduct from personal responsibility. Responsibility can be transferred to a superior, divided among many participants, concealed by sanitized language, or weakened by minimizing the consequences.
These mechanisms produce familiar explanations: the supervisor issued the order, the policy established the rule, the software generated the decision, or the employee completed only one assigned task. Each explanation removes the whole outcome from the responsibility of the person helping to produce it.
Milgram’s variations showed that obedience depended heavily on the surrounding structure. When the participant and learner occupied the same room, full obedience fell to 40 percent. Requiring the participant to force the learner’s hand onto a shock plate reduced it to 30 percent. Instructions delivered from another location lowered it to roughly one-fifth. Moving the experiment from Yale to an ordinary commercial office reduced full obedience to 47.5 percent.
The conduct of peers had the strongest effect. When two supposed participants refused to continue, only four of 40 participants, or 10 percent, reached the maximum level. The experimenter continued to issue the same commands, but those commands lost much of their force once refusal became visible and socially supported.
Jerry Burger’s 2009 partial replication recruited 70 adults and stopped at 150 volts to meet modern ethical standards. In the baseline condition, 70 percent were prepared to continue. Because the procedure ended early, it could not determine how many would have reached 450 volts. It did show that the pressures Milgram studied remained effective under modified conditions.
Later research has challenged the claim that participants simply surrendered their agency. Stephen Reicher, Alexander Haslam, and Joanne Smith argue that many participants identified with the experimenter and the stated purpose of the research. They continued because they believed they were contributing to legitimate scientific work.
Harmful institutions do not always need unthinking followers. They can recruit committed participants by presenting obedience as service to a valuable mission. The institution directs that commitment toward a narrow task, defines compliance as loyalty, and treats refusal as a failure of duty.
Roles Direct Attention and Incentives
The Stanford Prison Experiment is often used to claim that institutional roles automatically turn ordinary people into abusers. In August 1971, Philip Zimbardo and his colleagues placed 24 young men in a simulated prison in the basement of Stanford University’s psychology building. Participants were assigned to act as guards or prisoners. The planned two-week study ended after six days amid humiliation, distress, and abusive treatment.
The evidence does not support the simple conclusion that power naturally made the guards cruel and powerlessness made the prisoners passive.
Zimbardo served as both lead researcher and prison superintendent, compromising the distinction between observation and administration. The study had no neutral control group, no clear method for measuring behavioral change, and too few participants to support broad conclusions. Archival research by Thibault Le Texier found selective data collection, substantial direction from researchers, instructions about how guards should control prisoners, and clear signals about the conduct the experiment expected. The guards also behaved differently from one another; only some adopted aggressive behavior.
The experiment documents conduct inside one heavily directed environment. It does not prove that assigning a role automatically produces cruelty.
The 2002 BBC Prison Study offered a more conditional account. Stephen Reicher and Alexander Haslam placed 15 men in an eight-day simulated prison under stronger ethical safeguards. The guards did not automatically unite around their authority. The prisoners gradually formed a shared identity, challenged the hierarchy, and destabilized the original arrangement. Participants acted according to their roles when they identified with the group, its leadership, and the norms attached to it.
Research on deindividuation supports the same qualification. Anonymity does not simply remove restraint and release an antisocial instinct. It can make people more responsive to the norms of the group that matters most to them in that setting.
Roles shape conduct by directing attention, incentives, expectations, and permission. They tell participants what success means, whose judgment counts, and which consequences fall within their responsibility. A guard can be rewarded for maintaining order while ignoring dignity. A salesperson can be rewarded for meeting quotas while ignoring customer welfare. A bureaucrat can be evaluated for procedural accuracy while remaining insulated from the human result. A manager can learn to treat dissent as a threat to performance.
Harm becomes easier when the organization rewards it, gives it respectable language, limits the information available to each participant, and protects the conduct from scrutiny.
Institutions Turn Repeated Risk Into Normal Practice
The same mechanisms operate in real organizations, where the consequences are no longer simulated.
On January 28, 1986, the space shuttle Challenger disintegrated 73 seconds after launch, killing all seven crew members. The presidential commission identified the immediate cause as the failure of an O-ring seal in the right solid rocket booster. It also found organizational failures: engineering evidence did not reach the officials making the final decision, management judgment overrode serious concerns, and communication structures concealed the full level of risk.
Sociologist Diane Vaughan described the broader process as the “normalization of deviance.” Earlier anomalies had not caused disaster, so their repeated acceptance became evidence that continued acceptance was safe. An exception became a precedent, and the precedent became normal practice.
No single decision contained the entire catastrophe. Risk accumulated through technical classifications, launch schedules, management pressure, restricted communication, and repeated acceptance of a danger that had not yet proved fatal. Each decision could be defended within its immediate context even as the combined decisions produced a course no one could defend as a whole.
The Wells Fargo sales scandal shows how incentives and divided responsibility can produce the same pattern. A 2017 investigation commissioned by the company’s independent directors reviewed more than 35 million documents and conducted 100 interviews. It found that aggressive performance management, distorted sales incentives, decentralized control, deference to business leadership, and resistance to scrutiny pressured employees to sell unwanted products and, in some cases, open unauthorized accounts.
Approximately 5,300 employees were terminated for sales-practice violations between January 2011 and March 2016. Those terminations allowed the company to treat misconduct as a series of individual violations rather than confront the conditions producing it. Reports minimized the scale of the problem, while internal control functions lacked the independence and authority to challenge the sales organization.
The company rewarded sales, punished failure, distributed knowledge across departments, and isolated each violation from the larger pattern. Employees, managers, investigators, and executives could each describe a limited function. The institution produced systematic harm while its internal structure prevented anyone from being forced to accept responsibility for the complete result.
Compliance becomes self-reinforcing. Repetition makes a practice familiar. Familiarity reduces scrutiny. Continued practice then appears to confirm that the conduct is normal, accepted, and legitimate.
Making Resistance Possible
Telling individuals to be courageous does not remove the pressures that punish dissent. Resistance requires structures that reduce the cost of speaking, preserve responsibility, and prevent harmful decisions from disappearing into procedure.
Make disagreement visible early. Private doubt cannot correct a false public consensus. Objections must enter the record before silence is treated as agreement and before a decision gains momentum. Visible dissent gives others evidence that their concerns are shared.
Protect refusal in practice. A formal right to object means little when using it leads to dismissal, isolation, retaliation, or professional ruin. Independent reporting channels, enforceable anti-retaliation rules, whistleblower protections, and peer support reduce the price individuals pay for exposing harm.
Keep responsibility attached to decisions. Consequential decisions need identifiable authors, stated reasons, supporting evidence, and a record of objections. Long chains of command allow each participant to transfer responsibility elsewhere. A clear decision record shows who knew what, who decided, and who implemented the result.
Reduce distance from consequences. Decision-makers should encounter the people affected by their choices and receive direct evidence of the results. Harm is easier to authorize when those affected appear only as categories, cases, costs, or statistics.
Review the entire course of action. Independent checkpoints must examine the cumulative direction of a policy, not only the next incremental step. A decision can appear reasonable in isolation while extending a sequence that has become indefensible.
Separate oversight from performance pressure. Auditors, safety officers, investigators, and ethics functions cannot provide meaningful resistance when their budgets, access, or careers depend on the officials they must challenge. Oversight requires independent authority, adequate resources, access to decision-makers, and protection when its findings obstruct institutional goals.
Treat disruption as information. Systems organized around speed and performance often classify delay as failure and dissent as disloyalty. The employee challenging a metric, disputing approved language, or refusing an assignment may be identifying consequences the organization has trained everyone else to ignore.
These safeguards do not assume that every dissenter is correct. They allow objections to be examined without requiring a person to risk everything merely to make the evidence visible.
Organized Compliance Sustains Organized Power
Groups interpret uncertainty, authorities coordinate specialized knowledge, roles divide complex work, and institutions organize collective action. These capacities can support large-scale cooperation or large-scale harm.
A government cannot rule through direct force alone. It needs administrators to process decisions, professionals to legitimize them, workers to implement them, institutions to defend them, and a public that interprets widespread compliance as proof of consent. Corporations and other large organizations depend on the same distribution of participation.
This dependence establishes both the strength and the limit of organized power. Institutions can shape perception, assign roles, divide responsibility, reward cooperation, and punish refusal. They still require people to carry out their decisions.
Private disagreement leaves that capacity intact. Visible dissent breaks the appearance of unanimity and allows people to recognize that their doubts are shared. Protected, coordinated refusal goes further by interrupting the participation on which the system depends.
One dissenter can expose false consensus. Only organized refusal can stop organized harm.
Sources
Stanley Milgram, “Behavioral Study of Obedience”, Journal of Abnormal and Social Psychology 67, no. 4 (1963). ↩ ↩2
Solomon E. Asch, “Opinions and Social Pressure”, Scientific American 193, no. 5 (1955). ↩ ↩2
Axel Franzen and Sebastian Mader, “The Power of Social Influence: A Replication and Extension of the Asch Experiment”, PLOS ONE 18, no. 11 (2023). ↩
Rod Bond and Peter B. Smith, “Culture and Conformity: A Meta-Analysis of Studies Using Asch’s Line Judgment Task”, Psychological Bulletin 119, no. 1 (1996). ↩
Albert Bandura, “Moral Disengagement in the Perpetration of Inhumanities”, Personality and Social Psychology Review 3, no. 3 (1999). ↩
Stanley Milgram, Obedience to Authority: An Experimental View (Harper & Row, 1974); variations summarized in Nick O’Heron, “Obedience to Authority: Milgram”. ↩
Jerry M. Burger, “Replicating Milgram: Would People Still Obey Today?”, American Psychologist 64, no. 1 (2009). ↩
Stephen D. Reicher, S. Alexander Haslam, and Joanne R. Smith, “Working Toward the Experimenter”, Perspectives on Psychological Science 7, no. 4 (2012). ↩
Stanford University Libraries, Stanford Prison Experiment Archive. ↩
Thibault Le Texier, “Debunking the Stanford Prison Experiment”, American Psychologist 74, no. 7 (2019). ↩
Stephen Reicher and S. Alexander Haslam, “Rethinking the Psychology of Tyranny: The BBC Prison Study”, British Journal of Social Psychology 45, no. 1 (2006). ↩
Tom Postmes and Russell Spears, “Deindividuation and Antinormative Behavior: A Meta-Analysis”, Psychological Bulletin 123, no. 3 (1998). ↩
Presidential Commission on the Space Shuttle Challenger Accident, Report of the Presidential Commission, Volume I (1986). ↩
Diane Vaughan, The Challenger Launch Decision: Risky Technology, Culture, and Deviance at NASA (University of Chicago Press, 1996). ↩
Independent Directors of Wells Fargo & Company, Sales Practices Investigation Report (April 10, 2017). ↩
Empathic Philosophy begins with the human being rather than the state, market, party, institution, or doctrine. Its purpose is to establish a foundation for solidarity across political, cultural, and ideological divisions through recognition of our shared humanity.
Serious thought requires time, focus, and material support.
Paid subscriptions help sustain this publication and supports future projects.
Help carry the work forward with 50% off your first year.
This argument is meant to be tested. Add your perspective, challenge the analysis, or identify what it leaves unresolved.
Collective understanding develops through serious discussion. Join the subscriber chat to examine these ideas with readers committed to thinking clearly about power and what it will take to transform society.
#PokeTheEyeInTheSky
Enjoyed the listen at 1.50x and great content well written.