Predictably Irrational

In one of his best-known studies, Ariely asked a room of MBA students to write the last two digits of their Social Security number at the top of a page, next to a list of items — wine, chocolate, a cordless keyboard, a book. Then he asked them whether they'd pay that two-digit figure, treated as a dollar amount, for each item. Finally, he asked them to name the actual maximum they'd pay. The Social Security number is random. It has nothing to do with the value of a bottle of wine. And yet students with high digits bid substantially more than students with low digits — the top group offering sometimes two or three times what the bottom group would.

The random number became an anchor. Once it was in people's heads, it quietly set the frame for everything that followed, even though everyone, if asked, would have sworn their Social Security number had no bearing on what wine was worth. Ariely calls this arbitrary coherence. The first price we encounter may be arbitrary, but once it lands, our later judgments stay coherent with it. We don't know what things should cost in some absolute sense; we know what they cost relative to the last anchor we were handed.

Ariely's experiments on honesty are the ones that tend to stay with people. He gave participants a sheet of math problems, more than they could finish in the time allowed, and paid them per correct answer. One group handed their sheets to a proctor. Another group was told to score themselves, shred the evidence, and simply report how many they'd solved. The self-scoring group reported solving more — not wildly more, but consistently more. People cheated. What's interesting is how modestly they cheated. Nobody claimed to have aced the whole thing. They nudged their score up by a couple, took the small win, and kept their self-image intact.

That restraint is the discovery. We don't cheat as much as a cold calculation of risk and reward would predict, because we're also running a second ledger — the one that lets us keep thinking of ourselves as honest people. A small fudge survives that ledger. A big lie doesn't. So we skim a little off the top and stop before the number threatens our sense of who we are.

If the honesty studies show how fragile our self-control is under moral pressure, another strand of Ariely's work shows how completely it can dissolve under a different kind of heat. He and a colleague asked young men a series of questions about what they would and wouldn't do sexually, and about their willingness to behave recklessly or coercively to get what they wanted. The men answered twice: once in a calm state, and once while sexually aroused. The two sets of answers came from what looked like two different people. Aroused, the men reported willingness to do things that, calm, they had flatly ruled out.

The point is not that arousal makes people worse. It's that the calm self badly predicts the aroused self — and, crucially, doesn't know it. Each of us plans our life from the cool state, deciding what we'll do in the heat of a moment we can't actually feel yet. Ariely's finding is that the cool state has almost no access to the hot one. We are, in effect, strangers to the version of ourselves that shows up when the temperature rises, whether the heat is desire, anger, hunger, or fear.

The free Kiss, the Social Security anchor, the Ten Commandments, the two selves split by arousal — read together, they stop looking like a collection of odd glitches and start looking like a map. Each experiment locates a bias, then shows it firing the same way for almost everyone, under the same conditions. Ariely's wager is that this reliability is the good news hiding inside the bad. A random error can't be planned for. A predictable one can.