Would you say a line is longer when your own eyes say it isn't?

Not always, but often enough to surprise everyone. In 1951 the psychologist Solomon Asch put a college student in a room with seven other "students" and asked the group, one by one, which of three lines matched a line on a card. The answer was obvious. The catch was that the other seven were actors, instructed to give the same wrong answer on most rounds, and the real participant answered last.

Alone, people almost never got it wrong: the error rate was under 0.7 percent. With the unanimous actors in the room, about 36 percent of the real participants' answers copied the wrong majority. Yet the picture is not "everyone caves". Roughly two out of three answers stayed correct, and about a quarter of participants resisted every time. Asch himself wrote that it was a matter of concern that intelligent, well-meaning young people were willing to call white black.

Why did they do it? In interviews, some said they decided their own eyes must be wrong. Others knew the right answer and simply did not want to seem out of step.

The effect was also fragile. When just one actor gave the correct answer, only 5 percent of participants kept following the wrong majority. Answering in writing instead of out loud cut conformity too. A single ally, or a bit of privacy, was enough to loosen the group's grip.