Check your thinking

Each step was a classic experiment, cut down to a few seconds. Your lean on the wall came from your own answers. A few answers from one person cannot prove a bias. The studies below come from many people, and they do not all hold up equally well.

Why testing what fits feels like proof

Testing sets you expect to pass is not foolish. Klayman and Ha (1987) showed it is often a good strategy. It fails when your guess sits inside the real rule, the way “up by 2” sits inside “any numbers that go up”. Then every test you expect to pass does pass, and none of them can tell you that you are wrong.

What helps

Three habits push back. The first is to consider the opposite: before you settle, ask what would show you are wrong, then go and look for it. In the number rule, that means testing a set you expect to fail. In two experiments, telling people to consider the opposite cut bias more than telling them to be fair and unbiased (Lord, Lepper and Preston, 1984).

The second is a pre-mortem, the same move aimed at a plan. Imagine it is a year from now and the plan has failed, and write down why. Psychologist Gary Klein built the method to bring doubts out before a plan starts, not after (Harvard Business Review, 2007).

The third is to start from the reference class: before your own story, ask how long or how much it took others in the same spot. Students guessed their honors thesis would take about 34 days on average. It took 55.5 days (Buehler, Griffin and Ross, 1994).

Practice with feedback helps, a little, for a while. In two experiments, one game with personal feedback cut bias by 31.94% or more right away and 23.57% or more two months later; a video helped less (Morewedge and others, 2015). A 2025 review of 41 studies that taught students about biases found a small average gain (Swaryandini and others). What lasts is the habit: before you settle, ask what would prove you wrong.

What we left out, and why

Some famous ideas about the mind are not settled, so this room does not teach them as fact.

The Dunning-Kruger effect

The famous graph may be mostly a quirk of the statistics. With sounder tests on 929 people, Gignac and Zajenkowski (2020) found no sign of it, though a smaller effect may still exist for some skills.

Social priming

The idea that a passing cue, like a flag or a picture of money, quietly changes what people believe. In the 2014 rerun, flag priming and money priming did not replicate (Klein and others).

Ego depletion

Willpower as a tank that runs low. A test across 23 labs and 2,141 people found almost no effect (Hagger and others, 2016); a 2025 test across 14 samples with a much harder task found one (Dang and others). Unresolved.

The backfire effect

The fear that correcting a false belief makes people hold it harder. Across 52 issues and more than 10,100 people, no correction backfired (Wood and Porter).

Growth mindset hype

Short lessons promising big gains at school. Across 63 studies the average effect on grades was tiny and faded after correcting for bias (Macnamara and Burgoyne, 2023); another review found modest gains for the students expected to benefit most (Burnette and others, 2023).