Taking this research at face value, one might conclude that when people are not misjudging the world around them, they are lying to themselves about their own abilities and motivations. In one famous study, people were found to be "insensitive," beset by "ignorance," "general misconceptions," and a whole range of "shortcomings and biases." Krueger remembers a popular debate among social psychologists over which metaphor best drives home the depth of the mind's failings: Should researchers view the mind as a "cognitive miser," emphasizing our limited resources and reliance on irrelevant clues, or is the mind more accurately depicted as a "totalitarian ego," pursuing self-esteem at the cost of self-deception? Is your mind a Scrooge or a Stalin?

By the mid-1990s, Krueger began to wonder about the value of finding mistakes in human reasoning. His daughter was a toddler, and like many parents, he had become fascinated with her development. "I was overawed with the day-to-day advances in her thinking," he recalls. "What I was admiring was not her rational thought but her development of intuitive, associative, and automatic reasoning. In other words, I was admiring the same kind of thinking that social psychology researchers were finding fault with when they studied adults.

Was human reasoning really so flawed? Perhaps the errors lay in the means by which psychologists sought to explain them. Human thinking, Krueger notes, is of two broad types. There are the snap judgments we make on the fly, like assessing whether a person approaching us on the street is welcoming or threatening. And there are the activities to which we apply the full force of our minds, like preparing a business presentation or solving a math problem. That laborious reasoning has long been assumed to represent the gold standard of human thinking. It is the type of reasoning that social psychologists themselves employ. Test subjects, however, are typically placed in a situation and required to guess, react, or estimate. Later, the researcher analyzes the behavior at length, through the lens of statistics or logic. Whenever there is a disparity, the test subject is assumed to be displaying the error or bias, not the researcher.


Another problem with the studies, Krueger says, is that researchers are "null-hypothesis testing." Basically, they begin with the premise that the human mind is rational and then look for any deviation. Good behavior or moments of rationality are ignored because the intent is to study bad behavior. It's not unlike reality television: Unless there is some bad behavior, the research has nothing to show.

"I began to think that by comparing human judgment to objective reality, we were missing a bigger picture," says Krueger. "We were chronicling mistakes but stopping short of asking why such behavioral or cognitive tendencies existed or what general purpose they might serve. I began to think that bias and error couldn't be the end of the story."

The mind wants to believe that the line between good and bad behavior is clear. Looking again at the Milgram shock experiment, one wants to consider the subjects who administered "shocks" under order as cowards and those who refused as heroes. But imagine a different Milgram study. What if, when subjects showed up at the lab, they instead were confronted with smoke pouring out of the windows and a firefighter who told them, "Quick, help me carry this hose into this burning building." What would we think of those who followed authority in that situation? What would we think of those who refused?

It is an uncomfortable fact that the soldiers who ran the Nazi death camps and the soldiers who liberated them were all acting under orders from superiors. There is a world of difference in the moral implications of what they did, but the human tendency to obey authority resulted in both evil and good. Krueger's challenging question is this: Wouldn't scientists learn as much or more about mental mechanisms like obedience if they took its advantages into account? Couldn't we learn more about the bad by studying the good, or at least by examining bad and good behavior in the same context?

A rethinking of one particular classic of error research had a dramatic influence on Krueger's thinking. In a now famous study, Lee Ross and colleagues at Stanford University asked students if they would walk around campus wearing a sandwich board that read "Eat at Joe's." The test subjects who agreed to do this embarrassing task predicted that 62 percent of others approached to carry the sign would do it. But test subjects who refused to carry the sign thought that only 33 percent of others would agree to do it when asked. Researchers concluded that they had found a new bias in reasoning, which they called the "false consensus effect"—that people have the naive tendency to project their individual attitudes, values, and behaviors onto the majority.

Krueger was impressed by a critique of the study. Robyn Dawes, a professor Krueger had studied under, countered that the students who predicted that their opinions would be in the majority were not making an error at all but rather were taking their own opinion as a legitimate piece of data. "By definition, most people are in the majority most of the time," explains Krueger. "Therefore, if you assume that your opinion will match that of the majority, you will be right more often than not."

Krueger has taken this thinking a step further, to study the personal and social benefits of such behavior. In doing so, he may have cracked the "prisoner's dilemma," a classic experiment of both social psychology and economics. In the prisoner's dilemma, you are asked to imagine yourself alone in a cell, with an unseen companion isolated in a separate cell. You both are under suspicion of having committed a crime together, but the police don't have the evidence to convict you—yet. If you agree to betray your companion by testifying against him and he chooses to remain silent, you will be freed (zero years); if you both rat on each other, you receive a near-maximum sentence of three years. If you remain silent, and your companion does, too, you both receive only minimal time (one year), but if you stay quiet and your companion betrays you, you receive full punishment—five years, the sucker's outcome. Which choice, betrayal or silence, assures you the least time?