The Milgram Experiments

The man wired up in the next room has gone quiet. A minute ago he was screaming, pounding the wall, shouting that his heart was bothering him. Now, nothing. And the calm voice beside you says: “Please continue.” The next switch is yours to flip.

Most people flipped it. Two-thirds of perfectly ordinary adults obeyed all the way to a final lever marked 450 volts, “Danger: Severe,” and every single one went to at least 300. Beforehand, experts had bet that barely 1 in 100 would go that far (Milgram, 1963).

The terrifying finding was not that monsters will hurt people. It is that ordinary people will, when a calm voice in a lab coat says “continue.”



The setup

You are the “teacher.” Every time the “learner” (secretly an actor) gets a question wrong, you flip a switch for a bigger shock. The learner protests, screams at 150 volts, begs to be let out, then goes silent. Each time you hesitate, the experimenter says, evenly: “Please continue. The experiment requires that you continue.”

Most people kept going.

And here is the part that matters most: this was not cruelty. Participants sweated, trembled, argued, and begged to stop, and pressed the switch anyway. They did not want to. They did it regardless.


The cousin of Asch, one level up

  • Asch is sideways pressure: conforming to your equals.
  • Milgram is upward pressure: obeying an authority above you.

Milgram named the mechanism the agentic shift: you stop feeling like the author of your actions and start feeling like an instrument carrying out someone else’s will. Responsibility floats upward to the authority.

“I was just following orders” is not only an excuse. It is a real psychological state.


What turned obedience up or down

Milgram ran around 18 variations. The levers were clear:

LeverEffect
The authority’s presenceExperimenter in the room: 65%. Giving orders by phone: drops to ~21%. The authority has to feel present and legitimate.
Closeness of the victimThe more real and present the suffering, the less people obeyed.
A defiant peerWhen two other “teachers” refused, obedience crashed to 10%. One visible rebel frees everyone, exactly like Asch.

It still happens, and it is not mindless

Two findings kill the “that was just the 1960s” defense:

  • It replicates. A modern, ethically-revised version found virtually identical rates (Burger, 2009).
  • It is not blind. The newer engaged-followership reading shows people obeyed most when they identified with the cause (“this is important science”), and least when bluntly ordered. The single most coercive prod, “you have no choice, you must continue,” produced the most defiance.

People do not comply because they are forced. They comply because they are enrolled in something they believe is good.

The dangerous version is not a tyrant forcing you. It is a cause you believe in, asking for one more small step.


Why this matters

  • A uniform, a title, an institution does the heavy lifting. Legitimacy is granted, then exploited.
  • Harm escalates in tiny increments. You already went to 300 volts, what’s one more? Small steps slip past the alarm a single big ask would trigger.
  • Be, or find, the one who refuses. A single visible “no” collapses obedience for everyone watching.
  • Reclaim authorship. The question that breaks the agentic state is “who is actually responsible here, me or them?” The honest answer is always you.

Obedience does not feel like surrender. It feels like doing your job, which is exactly why it works.