A colleague gets promoted. Nothing dramatic changes. But six weeks later you notice he no longer says good morning to the security guard, and that he interrupts the junior people in a way he never used to.
The usual explanation is that power changed him. The more interesting claim is that power reveals character — that it did not install anything new, it just removed the reasons he had been behaving well.
Both claims have research behind them. They do not have the same amount of research behind them, and that gap is the most useful thing on this page. This is educational writing, not diagnosis or treatment.
Table of Contents
- Two Famous Claims, Pulling in Opposite Directions
- The Evidence That Power Reveals Character
- The Evidence That Power Changes You — and Why It Is Shakier
- So Which Is It?
- Why Removing Consequences Is the Whole Mechanism
- The Waiter Rule
- How to Read Someone Before They Get Power
- Frequently Asked Questions
- The Takeaway
Two Famous Claims, Pulling in Opposite Directions
Lord Acton wrote in 1887 that power tends to corrupt, and absolute power corrupts absolutely. The qualifier gets dropped in almost every repetition, which turns a careful observation into a law of nature.
The biographer Robert Caro, after decades spent documenting how Lyndon Johnson used power, arrived somewhere different: that power does not always corrupt, but it always reveals. Once someone no longer has to accommodate anyone, you finally find out what they wanted to do all along.
These are not compatible claims, and psychology has evidence pointing both ways. The question is not whether power reveals character or corrupts it — both appear in the literature. The question is which body of evidence is sturdier.
The Evidence That Power Reveals Character
The cleanest study here is Chen, Lee-Chai and Bargh, published in the Journal of Personality and Social Psychology.
They separated people by relationship orientation — roughly, whether someone approaches relationships communally (responsive to others’ needs) or as an exchange (tracking what is owed). Then they primed power, both with word cues and with environmental ones like sitting in a professor’s chair.
The result: power-primed communals behaved in more socially responsible ways, while power-primed exchangers acted more in their own self-interest. Same prime. Opposite behavior. The authors concluded that power activates whichever goals a person already associates with it — and that this happens nonconsciously.
That is the amplification finding in one sentence: power does not supply the motive. It supplies the permission.
The limitation: this is laboratory priming with student samples, and the wider social-priming literature has had serious replication problems as a field. The Person × Situation logic holds up better than any single effect size from it.
The Evidence That Power Changes You — and Why It Is Shakier
The other side is more famous, and this is where most articles stop.
Galinsky, Magee, Inesi and Gruenfeld ran four experiments plus a correlational study in Psychological Science. The best-known asked people to draw the letter E on their own forehead. High-power participants more often drew it facing themselves — readable to them, backwards to everyone else. Other experiments found high-power participants assumed others shared their private knowledge, and were less accurate at reading emotional expressions. Worth noting that perspective-taking and emotional concern are separable capacities, which is the same distinction that separates empathy from sympathy — a study can find a deficit in one without saying anything about the other.
It is a memorable study and it is cited constantly. Here is the part that almost never travels with it.
Many Labs 3 did not replicate an effect of power on perspective-taking using a sarcasm-interpretation task. A separate lab, testing emotion recognition, found the opposite of Galinsky’s result — power-primed participants made fewer errors. The original authors published a response in the Journal of Experimental Social Psychology in 2016 acknowledging the contested state of the evidence and arguing the effect is moderated rather than absent.
So the claim “power reduces empathy” is not fabricated. It is contested, dependent on which task you use, and considerably less settled than its popularity implies.
So Which Is It?
| Power reveals | Power corrupts | |
|---|---|---|
| Core claim | Amplifies existing dispositions | Reduces perspective-taking in anyone |
| Key evidence | Chen, Lee-Chai & Bargh (2001) | Galinsky et al. (2006) |
| Replication status | Person × Situation logic broadly supported | Failed in Many Labs 3; one lab found the reverse |
| Popular version | Under-quoted | Massively over-quoted |
| Honest summary | Better supported | Real but unsettled |
The defensible position: power reliably amplifies what a person already brings, and may also blunt perspective-taking in some people under some conditions.
Which means the popular formulation has it backwards. “Power corrupts” is the famous half and the weaker half. “Power reveals” is the quieter half and the sturdier one.
Dacher Keltner’s book-length treatment, The Power Paradox, sits in the middle of this debate and is worth reading — but a trade book making a case is a different kind of object from a replicated finding, and it should be weighted accordingly.
Why Removing Consequences Is the Whole Mechanism
Strip the jargon and the mechanism is unglamorous.
Most people are restrained less by their values than by their circumstances. You are polite to your manager because they decide things about your life. You do not interrupt in meetings because it would cost you. The behavior is real, and so is the cost keeping it in place.
Power removes the cost. Nothing gets added to the person — something gets subtracted from the environment. What remains is whatever they would have done if it had been free all along.
This is why the change often looks so fast. Nobody develops a new personality in six weeks. They stop paying a tax they had been paying invisibly for years.
It also explains a pattern people describe after getting something they chased for a long time: the relief of no longer needing to please anyone, and the discovery of who they are when the pressure is off. That overlaps with why people change after they get what they want, though arrival and authority are not the same event.
The Waiter Rule
The folk heuristic: watch how someone treats a waiter, a cleaner, a junior colleague — anyone who cannot help or harm them.
Two honest notes. There is no research validating the waiter rule as a predictive test. It circulates as executive folk wisdom, not as a finding. And it fails on anyone sophisticated enough to know they are being watched.
But it is a reasonable heuristic for a reason that follows directly from the amplification evidence: it samples behavior in the one setting where consequences are already absent. You are not predicting what power will do to them. You are observing a small-scale version of the thing itself.
A better version of the same test: watch what happens when they are slightly inconvenienced by someone with no leverage. A delayed order, a slow cashier, an assistant who made a small error. Kindness is cheap when nothing has gone wrong.
How to Read Someone Before They Get Power
None of these is proof. Together they are better than a gut feeling.
- How they talk about people who are absent — especially people below them.
- Whether their story has anyone else in it. Does their account of a success include the people who did the work?
- What happens when they are wrong in front of someone junior.
- Whether their politeness scales with usefulness. Warm upward, flat downward is the signal — warmth aimed at people who can confer something is closer to seeking approval than to kindness.
- How they behave with a small amount of authority now. Chairing a meeting, allocating something, going first.
- Whether they can be argued with by someone who cannot retaliate.
Number six does most of the work. Accepting a challenge from someone who has leverage over you is just self-interest. Accepting one from someone who does not is information.
And the uncomfortable inversion: everything above applies to you. The person best positioned to construct a flattering account of their own conduct is always you, which is the ordinary operation of self-deception rather than a character flaw.
Frequently Asked Questions
Does power reveal character or change it?
The evidence points more strongly toward revealing. Power priming has been shown to amplify what a person already associates with power — socially responsible people became more so, self-interested people became more self-interested. The claim that power itself reduces perspective-taking in everyone is real but contested, and a large multi-lab replication failed to reproduce it.
Is “power corrupts” scientifically accurate?
Only loosely. Lord Acton wrote that power tends to corrupt, and the qualifier matters. The experimental evidence for power reducing empathy exists but has replication problems; the evidence for power amplifying existing dispositions is sturdier.
Why do people change when they get promoted?
Often because a constraint was removed rather than because a trait appeared. Much everyday good behavior is maintained by consequences. Remove the consequences and what is left is what the person would have done anyway.
Does the waiter rule actually work?
There is no research validating it as a predictive test, and it fails on anyone who knows they are being observed. It is a reasonable heuristic because it samples behavior where consequences are already absent — which is the condition power creates.
Can someone with power stay decent?
Yes, and the amplification evidence is the reason to expect it. In the same studies where self-interested people got worse, communally oriented people got measurably better. Power is not uniformly corrosive; it is a multiplier with an unpredictable sign.
The Takeaway
Power reveals character the way a mirror does, not the way a mask does. It does not install ambition or cruelty; it removes the reasons for hiding them — and in people who were decent when it was expensive to be decent, it removes the reasons for hiding that too.
The honest version of the research is not the tidy one. Amplification has better support than corruption. The most quoted finding in this area is the least settled. And the people who repeat “power corrupts” with the most confidence are usually describing someone else.
Which leaves the only question that costs anything to answer: what would you stop doing tomorrow if there were no longer any consequence for stopping?
