Once a behavioral phenomenon has been identified in some experimental context, it is appropriate to start questioning its robustness. A popular and often productive questioning strategy might be called destructive testing, after a kindred technique in engineering. A proposed design is subjected to conditions intended to push it to and beyond its limits of viability. Such controlled destruction can clarify where it is to be trusted and why it works when it does. Applied to a behavioral phenomenon, this philosophy would promote research attempting to circumscribe the conditions for its observation and the psychological processes that must be evoked or controlled in order to eliminate it. Where the phenomenon is a judgmental bias, destructive testing takes the form of debiasing efforts. Destructive testing shows where a design fails; when a bias fails, the result is improved judgment.

The study of heuristics and biases might itself be seen as the application of destructive testing to the earlier hypothesis that people are competent intuitive statisticians. Casual observation suggests that people's judgment is generally “good enough” to let them make it through life without getting into too much trouble. Early studies (Peterson & Beach, 1967) supported this belief, indicating that, to a first approximation, people might be described as veridical observers and normative judges. Subsequent studies, represented in this volume, tested the accuracy of this approximation by looking at the limits of people's apparent successes. Could better judgment have made them richer or healthier?