Error Management Theory

Definition

One of the great challenges for humans is figuring out what is going on in other people's minds. People don't always disclose exactly what they are thinking, they can behave in very ambiguous ways, and sometimes they can be downright deceptive. For example, when a woman smiles at a man, is she sexually interested in him or just being nice?

Sometimes the errors people make in judging others are systematic, meaning that they tend to be biased in one direction or another. For example, judgments might be systematically biased toward a false positive error or a false negative error. In judging others, you would make a false positive error if you believed that a person had a particular thought or intention when the person actually did not. If you judged that the woman was sexually interested in the man, for instance, when she actually was not, you would make a false positive error. On the other hand, you would make a false negative error if you believed that a person did not have a particular thought or intention when the person actually did. If you judged that the woman was not sexually interested in the man when she actually was, you would make a false negative error.

Error management theory proposes that the direction of a bias in social judgment is tied to how costly different kinds of errors are. For example, consider how smoke alarms are designed. Failures to detect fires (false negative errors) are extremely costly, whereas false alarms (false positives) are usually just inconvenient. So, when engineers make smoke alarms, they tend to design them to be biased away from the more costly false negative error by setting a low threshold for fire detection. As a consequence, smoke alarms will tend to be systematically biased toward false positive errors (false alarms). A low threshold for fire detection will cause smoke alarms to make more errors overall, but it will minimize the cost of errors when they inevitably occur (i.e., the errors will tend to be false alarms rather than missed fires).

Error management theory proposes that the same principle of design applies to the evolution of judgment mechanisms in the human mind. Ancestrally, in many areas of social judgment, the costs of false positive and false negative errors differed. When the costs of false negatives are greater, error management theory predicts a bias toward false positives (as in the smoke alarm example); when the costs of false positives are greater, error management theory predicts a bias toward false negatives.

Examples and Evidence

One example of a false positive bias is in men's estimations of women's sexual interest. For an ancestral man, failing to detect sexual interest in a woman resulted in a missed reproductive opportunity, which was highly costly to his reproductive success. The opposite error (believing that a woman was interested when she was not) was perhaps a bit embarrassing but probably less costly overall. Thus, error management theory predicts that natural selection designed a bias in men toward slightly overestimating a woman's sexual [Page 312]interest to reduce the likelihood of a missed sexual opportunity; this leads modern men to overpercieve women's sexual interest. (The same prediction does not apply to women's perceptions because women need to invest very heavily in each offspring and because reproductive opportunities tend to be easier for women to acquire.) Evidence of this bias has been gathered in many types of studies. In laboratory studies of interactions between male and female strangers, men viewing the interaction tend to infer greater flirtatiousness in the female than do women viewing the interaction. In surveys of people's past experiences, women report more cases in which men overestimated their sexual interest than in which men underestimated it, whereas men's reports of women's overand underestimation errors do not differ. When men and women are shown romantic movies, men's subsequent tendency to see sexual interest in photographs of neutral female faces is greater than women's.

...