Anger as a commitment device

Why can visibly losing your temper win a better deal than staying calm and reasonable?

We are taught that anger in a negotiation is a failure of self-control — that the composed party has the advantage, because they can still see their options clearly. And yet the person who visibly loses their temper over a small sum quite often walks away with more than the person who reasons patiently for the same thing.

That is odd enough to be worth stopping on. If anger only clouded judgement, it should cost the angry party, not the calm one. So either the observation is an illusion, or losing an option can be worth something. Which of those seems more promising to you?

Take the second and see where it leads. Consider a simple split: I propose how to divide a hundred pounds, you accept or refuse, and if you refuse we both get nothing. Suppose we are both coldly rational and both know it. I offer you one pound. Refusing costs you a pound and gains you nothing, so you accept. Knowing you will accept, I never offer more. Being reasonable has made you cheap.

Now suppose you announce beforehand: offer me less than forty and I will refuse. If I believed you, I would offer forty. Do I believe you? Only if refusing would actually be in your interest at the moment of choosing — and it is not. The threat is empty precisely because you remain free to act sensibly when the time comes. This is Thomas Schelling's insight from The Strategy of Conflict in 1960: in a bargain, the power lies not with the party who has the most options but with the party who can credibly give options up.

So what could give the threat teeth? Something that changes your payoffs so that refusing is no longer the worse choice for you — or something that removes your ability to reconsider. And notice what anger does. It makes refusal feel good. The insult now carries a cost of its own, so accepting a humiliating offer is no longer costless, and carrying out the threat is no longer irrational.

Now the second half, which is the one people usually miss: how do I know your anger is real and not a performance? Here the involuntary quality of the display starts to look less like a design flaw and more like the point. A flushed face, a shaking voice, a raised pitch — these are hard to produce convincingly on demand and hard to suppress when genuinely felt. A signal that is expensive or difficult to fake is exactly the kind that can be trusted. Robert Frank's Passions Within Reason in 1988 built this argument out: the emotions that seem to undermine our interests may be there to make our commitments believable.

And behaviour matches the prediction well enough to be interesting. In ultimatum experiments across many populations, low offers are routinely refused at real cost to the refuser, and proposers anticipate this and offer far more than the rational-actor account predicts. On the emotional side, Sell, Tooby and Cosmides argued in 2009 that anger functions as a bargaining system for adjusting how much weight others give your interests, and reported that people better positioned to impose costs — physically stronger men in their data — both anger more readily and expect better treatment.

Two honesties. This is a functional account of what anger is for, and functional accounts in this area are argued over, not settled. And a device that works by removing your ability to back down will keep working when backing down was the right move — so the same mechanism that wins the bargain also produces the feud and the ruined friendship.

Note what the argument does not claim. It does not say anger is a strategy, in the sense of a plan someone forms. Quite the opposite: if it were consciously chosen it would be revocable, and a revocable commitment is no commitment. The strategic value depends on the person not being in charge of it. That is the resolution of the paradox we started with — the loss of control is not a side effect that the mechanism survives, it is the mechanism.

It also does not claim that anger is good for the angry person. A commitment device is deliberately a way of tying your own hands, and hands stay tied whether or not that turns out well. The person who refuses a derisory offer has lost real money to gain a reputation and to deter the next such offer. Whether that trade pays depends entirely on whether there will be a next time and whether anyone is watching — which is why the account predicts anger should be most useful in repeated dealings among people who know each other, and least useful with strangers you will never see again.

One misconception to correct gently: that this makes anger a bluff. A bluff is a false signal that works while it is believed. This is nearer the reverse — the display works because it is not false, and its usefulness would collapse if it could be produced at will by anyone.

• Why anger becomes worthless in a group that has learned to discount it, and what that implies about who can afford to use it.
• Whether the same logic explains guilt and gratitude, which Frank treats as commitments to keep bargains rather than to break them.
• How cultures that strongly discourage displays of temper substitute other credible commitments.

Every term the collection defines is gathered in the glossary.