Judgment Under Uncertainty: Heuristics and Biases

The book argues that people rely on a limited set of cognitive heuristics—representativeness, availability, and anchoring—when making judgments under uncertainty, and that these heuristics produce systematic and predictable biases rather than random errors. This directly challenges the rational-agent model of human judgment, which assumes people intuitively apply correct statistical principles such as Bayes' theorem. The volume's opening chapter, Tversky and Kahneman's 1974 Science paper, establishes this framework by demonstrating that judgments of probability are governed by similarity assessments and ease of recall rather than by proper probabilistic reasoning. The subsequent chapters extend this claim across domains including medical diagnosis, covariation assessment, overconfidence, and risk perception.

Judgment Under Uncertainty is a technical anthology of primary research papers, roughly half previously published and half newly written, intended for researchers already familiar with experimental psychology and probability theory. Thinking, Fast and Slow, published nearly three decades later, is a synthetic trade book that reorganizes the same research program into a unified narrative built around the System 1/System 2 distinction. The later book simplifies and popularizes; the earlier volume documents, with full experimental detail and statistical analysis, the specific studies that ground the broader claims. Kahneman has described the earlier collection as the scholarly foundation upon which the later synthesis was built.

The most sustained critique comes from Gerd Gigerenzer, who argued in a series of publications beginning in the 1990s that the program imposed unnecessarily narrow norms of reasoning. His core claims: first, that many biases diminish or disappear when problems are framed in terms of frequencies rather than single-event probabilities; second, that vague heuristic labels like 'representativeness' explain everything and nothing because they lack algorithmic specificity; and third, that characterizing judgments as 'errors' presupposes a normative standard that may be inappropriate for single-case judgments. Kahneman and Tversky responded directly in a 1996 Psychological Review exchange, defending both the empirical robustness of their findings and the appropriateness of probabilistic norms for evaluating judgment quality.

The representativeness heuristic, which leads to base-rate neglect and the conjunction fallacy (the 'Linda problem'), remains among the most widely cited concepts. The availability heuristic, linking judgments of frequency to ease of recall, has been applied extensively in risk perception research, explaining why vivid, widely publicized dangers are overestimated relative to mundane but statistically more common ones. Anchoring and adjustment has proven remarkably robust and is now a staple in negotiation research and behavioral economics. The simulation heuristic, the illusion of control, and hindsight bias, also featured in the volume, have each generated substantial literatures. The volume's chapters on overconfidence and calibration established the empirical foundation for subsequent work on expert judgment.

A frequent oversimplification treats the program as demonstrating that human beings are simply irrational or fundamentally flawed in their reasoning. The volume's editors and contributors explicitly frame heuristics as generally adaptive and efficient mechanisms that work well in most everyday contexts but fail in specific, identifiable situations—particularly those involving statistical inference. The opening chapter describes heuristics as useful because they reduce complex judgment tasks to simpler operations, and the final sections of the book include chapters on debiasing and improving judgment, indicating that the program's aim is diagnosis and correction, not condemnation of human cognition as such. The biases are systematic, not pervasive.

Behavioral economics was directly shaped by the volume's documentation of judgment biases, which provided empirical grounding for challenges to expected utility theory. Medical decision-making research adopted the findings on probabilistic reasoning and base-rate neglect to study diagnostic errors and improve clinical training. Legal scholarship has drawn on the work on hindsight bias and anchoring to examine judicial reasoning and jury behavior. Public policy applications include the 'nudge' framework developed by Thaler and Sunstein, which uses insights about systematic biases to design choice architectures that improve outcomes without restricting options. Risk communication and intelligence analysis have also incorporated findings on availability and overconfidence.