The Representativeness Heuristic

Judging by How Things Look

The representativeness heuristic is the tendency to judge the probability of something based on how closely it resembles a prototype, rather than on actual statistical likelihood. Kahneman and Tversky demonstrated this with their famous 'Linda problem': told that Linda is a philosophy graduate concerned about social justice, most people judge it more likely that she is 'a bank teller AND a feminist activist' than 'a bank teller' — even though the conjunction of two events is always less probable than either event alone.

People make this error because 'feminist bank teller' resembles the description of Linda more than 'bank teller' alone does. We substitute the question 'what is probable?' with 'what is representative?' — and the two are not the same thing.

Base Rate Neglect

The representativeness heuristic leads to a specific error called base rate neglect — ignoring the underlying probability (base rate) in favor of descriptive information. If someone tells you they met a quiet, organized person who loves puzzles, you might guess 'librarian.' But there are far more accountants, teachers, and office workers in the population than librarians. The description 'fits' librarian, but the base rate makes almost any other profession more likely.

In security contexts, base rate neglect is dangerous. After a terrorist attack, people who 'look like' the attackers face suspicion, even though the vast majority of people fitting any profile are not terrorists. The base rate of terrorism is extremely low, but the representativeness heuristic causes people to overweight resemblance to the prototype.

In medicine, the same error occurs. A patient with symptoms that resemble a rare disease may be diagnosed with that disease — even when a common condition is far more likely given the base rates. As the medical saying goes: 'When you hear hoofbeats, think horses, not zebras.'