How To Disagree Without Dehumanizing

In 1994, Radio Mille Collines broadcast one phrase over and over in Rwanda: "cut down the tall trees." The Tutsi were the tall trees. Within a hundred days, 800,000 people were dead — most of them killed not by soldiers with guns but by neighbors with machetes. People who had borrowed tools from each other, attended each other's weddings, eaten from the same pot.

The language came first. It always does.

Philip Zimbardo, who ran the Stanford Prison Experiment and later devoted years to studying how ordinary people become capable of atrocity, identified dehumanization as the central mechanism. You don't need a monster. You need a process. And that process begins the moment a human being stops being a person in your mind and starts being a symbol of everything you fear or hate.

The psychologist Nick Haslam calls this "infrahumanization" — the subtle, often unconscious process by which we deny full human complexity to outgroup members. We grant our ingroup the full range of human emotion: ambivalence, complexity, dignity, the capacity to be right and wrong simultaneously. We grant the outgroup a reduced set: predictable, one-dimensional, driven by base motives. They're angry. They're greedy. They're naive. They're dangerous. They're ridiculous.

You don't have to be a Nazi to do this. You just have to be human, under stress, in conflict, with a strong moral conviction and an outgroup to project it onto.

This is not a fringe phenomenon. This is Tuesday.

Here's the uncomfortable inversion: the more righteously correct you believe your position to be, the more license you feel to dehumanize the opposition.

Jonathan Haidt's research on moral psychology shows that we reason about morality the way a lawyer argues a case — backwards. We feel first, then we construct the justification. This means that when someone holds a position that triggers your moral disgust response, you experience your subsequent contempt for them as principled. It feels like clarity. It feels like seeing clearly while others are blinded.

It is not clarity. It's the brain's reward system giving you a hit of righteousness and calling it wisdom.

This is why "punching up vs. punching down" arguments — while structurally useful in some contexts — can become permission structures for dehumanization when applied carelessly. Yes, power dynamics matter. Yes, not all contempt is equivalent. But the mechanism of dehumanization doesn't care which direction it flows. It corrodes the person using it, poisons the group watching, and makes resolution impossible.

The history of revolutions is a history of liberation movements that, upon gaining power, reproduced the same dehumanization they were born fighting. Not because the people involved were uniquely corrupt, but because they used the tool. And the tool is what it is regardless of who swings it.

The essayist Paul Graham published what he called a "hierarchy of disagreement" — a scale ranging from the lowest (name-calling) to the highest (directly refuting the central point of an argument). It's worth naming the full ladder, because most people believe they operate at the top while they're actually camped at the bottom.

Level 7 — Refuting the central point: You identify the strongest version of the other person's argument and show why it's wrong. This is rare. It requires you to understand their position better than they stated it.

Level 6 — Refutation: You contradict what they said and back it up with evidence and reasoning.

Level 5 — Counterargument: You make an argument against their position, though not necessarily against their specific claim.

Level 4 — Contradiction: You simply assert the opposite. No evidence, just opposition.

Level 3 — Responding to tone: You attack how they said something rather than what they said. "That's really arrogant." "You sound defensive."

Level 2 — Ad hominem: You attack the person's character or circumstances to discredit them rather than engaging the argument. "Of course you'd say that — you're one of them."

Level 1 — Name-calling: The argument is entirely absent. What's left is a label.

The majority of political discourse operates between levels 1 and 3. The majority of online argument never leaves level 1. And none of these lower levels change anyone's mind — they harden it. Every time you attack someone at level 1, you make them more certain they're right, more committed to their tribe, less open to the possibility that they've missed something.

This is not a theoretical concern. It's measurable. Brendan Nyhan's research at Dartmouth showed that corrections to factually incorrect beliefs frequently backfire — respondents exposed to corrections became more convinced of their original (wrong) belief. The correction triggered identity defense. The person stopped hearing evidence and started defending territory.

The implication is brutal: if you actually want to change something, contempt is the least effective tool available to you.

One of the most consistent findings in conflict psychology is that people in conflict attribute their own side's behavior to noble motives and the other side's behavior to ignoble ones. We protest because we care. They protest because they're manipulated. We govern from principle. They govern from corruption. We have concerns. They have agendas.

Adam Galinsky at Columbia calls this the "motive attribution asymmetry" — and it's nearly universal across political, ethnic, religious, and organizational conflicts. It produces a specific kind of stalemate: both sides are certain the other is acting in bad faith, so no gesture from the other side is ever taken at face value.

This doesn't mean both sides are equally right. It doesn't mean there's always a "both sides" to have. It means that if you want to have a real conversation — one that could produce genuine understanding, genuine persuasion, or genuine change — you have to suspend the motive attribution long enough to actually hear what the other person is saying.

Not forever. Not unconditionally. Just long enough to actually understand what you're arguing against.

Before the conversation:

Know what you're actually trying to accomplish. Are you trying to win? Are you trying to understand? Are you trying to persuade? Are you trying to maintain relationship while holding your ground? These are different goals and they require different approaches. Most people enter arguments with no clear goal and then wonder why the argument went nowhere.

Steelman before you engage. This means constructing the strongest possible version of the other person's argument — not the version that's easiest to dismiss. Ask yourself: what would a smart, well-intentioned person have to believe for this position to make sense? This is not the same as agreeing. It's the intellectual equivalent of knowing your opponent's best moves before the match starts.

Locate your own stakes. What are you actually afraid of losing if the other person is right? This is not comfortable to sit with. But if you don't know your own emotional investment, it will run the conversation without your permission.

During the conversation:

Attack the idea, never the person. This sounds obvious. It isn't practiced. "That policy would cause serious harm for these reasons" is an attack on the policy. "You only believe that because you're privileged/brainwashed/naive/corrupt" is an attack on the person. The first can be argued with. The second only produces defensiveness and entrenchment.

Use "and" not "but." When you acknowledge something true in the other person's position, don't immediately cancel it with "but." "I hear what you're saying, but..." is the verbal equivalent of saying "I hear nothing you're saying." Try "I hear what you're saying, and here's where I see it differently." The "and" keeps both realities in the room at the same time.

Name the process when it goes bad. If the conversation starts sliding into dehumanization — from either direction — you can name it without accusing. "I notice we're starting to talk about this in terms of what kind of people believe what, rather than the actual argument. Can we go back to the specific question?" This is not weakness. It's the move of someone who actually wants to get somewhere.

Give them an exit ramp. People rarely change their minds in the moment. They do it later, privately, when they're not under threat of looking weak in front of witnesses. If you make it socially impossible for someone to concede a point, they won't. Design the conversation so they can shift without losing face.

After the conversation:

Don't celebrate if you "won." If someone walked away feeling humiliated, you didn't advance the cause — you recruited another soldier to the other side. The goal was never to defeat a person. It was to change a mind, or failing that, to model a different way of having the fight.

Ask what you learned. Even — especially — in arguments with people you believe are badly wrong, there is something to learn. Either about the reasons people adopt a position, or about where your own argument has gaps, or about what values you didn't know you held. Every conversation is data.

A single conversation between two people who disagree is low stakes. What happens when a society dehumanizes at scale is not.

The psychological research on group behavior shows a consistent pattern: once dehumanizing language becomes normalized within a group, it creates permission for escalating acts. The language doesn't just reflect contempt — it produces behavior. You can trace the linguistic shifts in media coverage, political speech, and public discourse in the months leading up to every major ethnic conflict of the last century.

This is why the way you argue isn't separate from what you're arguing about. If you're arguing for human dignity using methods that strip dignity from your opponents, you're teaching the lesson your methods carry. If you're arguing for a better world using the logic of dehumanization, you're building the world your methods create.

The people who dismantled apartheid in South Africa didn't just need the right policies. They needed people who could be in the same room as their oppressors and treat them as human beings capable of change. Nelson Mandela spent 27 years in prison and walked out and shook hands with the architects of his imprisonment — not because he had stopped believing they were wrong, but because he understood that the future he wanted couldn't be built on a foundation of righteous contempt.

This is what it costs. It costs exactly that much.

These are not scripts. They're habits of mind that can be built with repetition.

1. Separate the idea from the person. In every argument, hold two things separately: "this person is fully human and deserves basic respect" and "this specific position is wrong and I intend to say so clearly." Both can be true. They are almost always both true.

2. Ask what they're protecting. People hold positions because those positions protect something they value — security, identity, community, family, a sense of justice. If you can identify what someone is protecting, you understand the argument at a deeper level. You may still disagree completely, but you're arguing with the real thing.

3. Distinguish between person and position in your language. "That argument is wrong" rather than "you're the kind of person who would say that." "I disagree with that decision" rather than "of course you'd do that." The specific words matter. They train your own mind as much as they signal anything to the other person.

4. Stay curious longer than is comfortable. Curiosity and contempt cannot occupy the same moment. You can choose which one to stay in. Curiosity doesn't mean approval. It means you're committed to understanding what you're actually dealing with rather than the version of it that's easiest to dismiss.

5. Notice when you're arguing to win vs. arguing to understand. These are different modes. Sometimes arguing to win is appropriate — in formal debate, in legal proceedings, in advocacy contexts. But in most human conversations, the goal that actually serves everyone is understanding. When you're trying to understand, you ask more questions and make fewer declarations.

6. Hold the long view. The person across from you today is not your enemy for eternity. They are a person who holds a position you find wrong, in a moment in history, in a context that shaped them. Positions change. People change. The way you treat someone in a hard conversation is something they remember long after the specific argument is forgotten.

Law 0 is "You Are Human." Not "You Are Right." Not "You Are Good." Human — which includes the full range: capable of great harm, capable of great repair, always more complex than the worst thing you've done or the worst position you've held.

Disagreement is one of the primary sites where Law 0 gets violated. When you dehumanize someone you disagree with, you're not just being tactically ineffective. You're betraying the foundational law. You're saying: your humanity is conditional on your agreement with me.

That's not a political position. That's the mechanism of every atrocity ever committed.

The inverse is equally true. When you hold someone's humanity — even as you fight them on the argument — you're practicing the foundational law in one of its hardest forms. You're saying: your humanity is not contingent on your being right. You belong to the species even when you're badly wrong.

This is what changes the world. Not the winning. Not the righteous contempt. The willingness to stay in the argument with someone you believe is badly, seriously, maybe dangerously wrong — and treat them like a person the whole time.

It is the hardest thing. It is also the only thing that works.