Situations Beat Character

Here is the claim in its strongest form, the one you have probably heard:

Put a normal person in the right situation and they will do almost anything. Character barely matters. The situation decides.

You may already know the two experiments usually offered as proof. This lesson is going to do something a little unusual with them. We are going to look at the studies, then look carefully at what is wrong with them, and then see what is left standing.

That order is deliberate. This course made a promise in Lesson 2.4 about showing you contested evidence as contested. This is the lesson where that promise costs something, because the tidy version of this material is much more persuasive than the accurate version, and the tidy version is what almost every book, documentary, and psychology class will hand you.

The Two Famous Studies

The obedience experiments. Beginning in 1961 at Yale, a researcher named Stanley Milgram ran a series of studies in which volunteers were told they were helping with an experiment on learning. A volunteer sat at a machine of switches and was told to deliver an electric shock to a learner in another room each time the learner made a mistake, increasing the shock by a small amount each time. The switches ran up to a level marked as extremely dangerous. The learner was an actor. There were no shocks. As the levels rose, the learner protested, then screamed, then stopped responding entirely. When the volunteer hesitated, a man in a lab coat told them to continue.

In the version everybody quotes, about two thirds of volunteers went all the way to the end of the board.

The prison study. In 1971 at Stanford, Philip Zimbardo assigned college students to be guards or prisoners in a simulated jail in a basement. The study was planned for two weeks. It was stopped after six days, reportedly because the guards had become abusive and the prisoners had broken down.

Together these produced the story that filled textbooks for fifty years: ordinary people, assigned a role, become cruel. Anyone can be a torturer. The situation is everything.

What Is Wrong With Them

The prison study should not be treated as evidence at all.

This is a strong statement and it is now well supported. Over the last decade, researchers have gone through the study's own archived recordings and papers, and interviewed surviving participants, and the picture that came out is not what the story described.

The guards were not left to develop cruelty on their own. Recordings show the research team coaching them toward harshness, instructing them on the attitude required and pressing individual guards who were not being tough enough. The most famous prisoner breakdown, the one that appears in every retelling, was later described by the participant himself as performance. He wanted out, said so, and found that the fastest exit was to fall apart convincingly.

There are structural problems on top of that. Participants were recruited with a newspaper ad that specified a study of prison life, and a later study found that people who answer that ad, as opposed to an otherwise identical ad without the word prison, score meaningfully higher on aggression and authoritarianism and lower on empathy. So the sample was pre-sorted. There was no control group. There were about two dozen participants. And Zimbardo himself played the prison superintendent, which means the person running the study was a character inside it.

You may still find this study taught as fact. It is a demonstration, and a piece of theater, and it belongs in the history of the field rather than in its evidence base. When you next see it cited as proof that anyone can become a torturer, you now know something the person citing it does not.

The obedience experiments are damaged but not destroyed.

The problems here are real and smaller.

The famous two thirds figure comes from one variation out of roughly two dozen that Milgram ran, and across those variations the compliance rate ranged from nearly everybody to almost nobody. Presenting one number as the result of the experiment is misleading, and Milgram himself contributed to that by featuring his most dramatic condition.

Archival work has also shown that the experimenter's script was not followed as advertised. The four standardized prods were supposed to be delivered in order and then the session ended. In practice, some volunteers were pressed many more times than the protocol allowed, with improvised pressure well beyond the script.

And a meaningful number of participants said afterward that they had suspected the shocks were not real. Uncomfortably, the ones who suspected were more likely to keep going. That does not erase the finding, but it means the number of people who knowingly inflicted serious pain on a stranger is smaller than the headline suggests.

What survives is still substantial. A careful partial replication in 2009, stopping at a lower point on the board for ethical reasons, found compliance rates statistically indistinguishable from Milgram's. Similar patterns have appeared in other countries and other formats. Something real is being measured. A large fraction of ordinary people will continue to do something they believe is seriously harming another person when an authority figure asks them to continue.

The Reinterpretation That Changes Everything

Here is the part that is more interesting than the original finding, and much less well known.

If you go back to Milgram's recordings and read what participants actually said, they do not sound like people blindly following orders. They sound like people who have been persuaded.

They ask about the purpose of the study. They express concern for the learner and then accept reassurance. They are worried about ruining the data. Several of them are, in a real sense, participating in science, and they continue because they have come to believe the science matters and that a responsible person sees it through.

The strongest evidence for this is in the prods. Milgram's four prompts escalated in force. The first three appealed to the experiment: please continue, the experiment requires that you continue, it is absolutely essential that you continue. The fourth was a flat order: you have no other choice, you must go on.

The fourth one did not work. When the experimenter resorted to naked command, participants overwhelmingly stopped.

Think about what that means. The prompt that sounds most like obedience was the one that produced refusal. What kept people going was not being ordered around. It was identification with something they believed was worthwhile, delivered by someone they saw as a legitimate representative of it.

This is a considerably more disturbing finding than the original. It means the lever is not authority. It is belonging, and purpose, and the wish to be a good contributor to something that matters. And it lines up exactly with Lesson 3.1: the people doing harm believe they are the good ones, and they are not being compelled, they are being enlisted.

Module Six is about that lever. This is where it first appears.

What Actually Survives

So after all that, what can you rely on? Five things, all better supported than the famous headline.

One: gradualness. The board went up in small steps. At no point was any single step a large increase over the one before, and there was never an obvious place where a person could say that this is where it becomes wrong. This is not a laboratory artifact. Financial fraud, escalating force, abusive relationships, and institutional corruption almost all proceed by increments that are individually defensible. The first step is the only one that presents itself as a decision. Every step after it is a small extension of a position you have already taken.

Two: distance. This is among the most consistent results in the whole set. Compliance dropped substantially when the learner was in the same room, and dropped further still when the volunteer had to physically place the learner's hand on a plate. It also dropped sharply when the experimenter left the room and gave instructions by telephone.

Both directions matter. Distance from the consequence makes harm easier. Distance from the authority makes refusal easier. Every large organization that has ever done damage has been built, deliberately or not, to maximize the first and minimize the second.

Three: somebody else is responsible. Volunteers frequently asked who would be held accountable, and the assurance that the experimenter accepted responsibility mattered a great deal.

Scale that up and you have how institutions do harm. Divide any action among enough people and nobody performs the whole act. One person writes the policy, another approves the budget, another files the form, another drives the vehicle. Each has a defensible fragment. The act exists only in the aggregate, where no one lives.

A note of caution on the classic version of this, since it comes up: the famous story about the murder in New York where dozens of witnesses supposedly watched and did nothing turns out to have been substantially wrong in its details, and several people did act. The diffusion effect is real in laboratory settings, but a large study of actual recorded street conflicts found that somebody intervened in the great majority of them. People are more likely to help than the folklore says. Diffusion of responsibility is a real force, and it is not the whole story about human beings.

Four: roles, but not in the way you were told. Since the prison study will not support the weight put on it, is there anything better? Yes. In 2002 two British researchers ran a much more carefully designed prison simulation, with real safeguards, independent ethical oversight, and it was recorded and broadcast.

What happened was not what the earlier study predicted. The guards did not spontaneously become tyrants. In fact they were uncomfortable with their authority and failed to cohere as a group. The prisoners, who did cohere, effectively took over. But the story did not end well. The resulting egalitarian arrangement was unable to function, and as it collapsed, participants from both groups began proposing a harsh, authoritarian order, which is when the study was ended.

That is a better and more usable lesson than the original. Cruelty did not come from a uniform. It came from a group that was failing, and the appetite for an authoritarian solution grew out of that failure. Tyranny is more often invited by breakdown than imposed by costume.

Five, and most important: one other person changes everything. In the variation where the volunteer worked alongside two others who refused and walked out, compliance nearly collapsed. In an entirely separate line of research on conformity, where people were asked to give obviously wrong answers about the lengths of lines because everyone around them had, a single dissenting ally cut conformity dramatically.

Sit with the size of that effect. Not a persuasive speech. Not a moral argument. One other person, visibly not going along.

This is the most practically useful finding in this module, and it is the one Lesson 1.4 was pointing at. When you hesitate visibly, you are not just removing yourself. You are altering the situation everyone else is standing in. You become, for the person next to you, the thing that made refusal thinkable.

The Correction To The Correction

The slogan overshoots, and the overshoot has done damage of its own.

In every one of these studies, a minority refused. In the best known obedience condition, a third of people stopped, some early, some furiously. In every genocide on record, there were rescuers. The claim that situations override character is not what the evidence shows.

What the evidence shows is that situations reliably shift the distribution, and do not determine the individual. A situation can take a population where almost nobody would harm a stranger and produce one where most people will. It cannot tell you which ones.

And when researchers have gone looking for what distinguishes the refusers, the answer is less inspiring than anyone wants. Interviews with people who sheltered others at enormous personal risk during the Holocaust turned up remarkably little moral eloquence. Very few described a decision. Many said some version of: they were at my door, what else could I do. Their explanations tended to be flat, practical, and unheroic, and a striking number involved simply having been asked directly by someone standing in front of them.

Bear in mind that this comes from interviews conducted decades afterward, so people are reconstructing. But the pattern is consistent enough to take seriously, and what it suggests is that refusal is less about moral quality and more about a handful of concrete conditions: whether the situation was concrete rather than abstract, whether the person had a prior commitment that already covered the case, and whether anybody else visible was refusing.

What You Can Actually Do

This is where Module One's second claim gets specific. Being a less reliable recruit is not a matter of moral fiber. It is a matter of five ordinary, unglamorous habits.

Watch the first step. It is the only one that will announce itself. Everything after it will feel like consistency, and consistency is the strongest pull there is.

Decide in advance and say it out loud. Write the sentence before you need it. Not a value. A behavior, in a context you actually occupy, specific enough that you would recognize the moment. A decision made in advance is much cheaper than one made under pressure, because under pressure you are not deciding, you are defending a position you have already taken.

Be the one who speaks first, in small things. You will almost never face a dramatic test. You will face many small ones: a meeting, a group chat, a family dinner, a form. Practice there, on things that cost you almost nothing, because the capacity is trainable and it does not appear on demand.

Keep your name on things. Whenever an action has your name attached, the diffusion mechanism cannot run. Ask who is accountable, out loud, and notice how the temperature of a room changes when somebody does.

Stay close to the consequence. If you cannot see who is affected by what you are part of, go and look. Distance is the single most reliable enabler in the whole set of findings, and it is usually the easiest one to remove.

None of that is heroic. That is the point. Heroism is not a plan. These are.