View source for Heuristics and Biases

'''Heuristics and Biases''' is the name of a research program in cognitive psychology, launched by [[Daniel Kahneman]] and [[Amos Tversky]] in the early 1970s, that transformed how we understand human judgment under uncertainty. The program's central claim is that humans rely on [[Mental Heuristics|mental heuristics]] — fast, computationally cheap cognitive shortcuts — to navigate a world of incomplete information and time pressure. These heuristics are adaptive but systematically produce predictable deviations from the norms of [[Expected Utility Theory|expected utility theory]] and [[Bayesian Probability|Bayesian probability]]. The program is empirically one of the most robust bodies of work in psychology; conceptually, it is one of the most contested.

The name itself encodes a tension. '''Heuristics''' are the mechanisms; '''biases''' are the errors they produce when deployed outside their ecological competence. The program studied both, but popular reception has focused overwhelmingly on the biases — turning a theory of adaptive intelligence into a catalog of human stupidity. This distortion is itself a bias: the [[Availability Heuristic|availability heuristic]] operating on the most vivid and counter-intuitive findings of the program.

== The Three Anchor Heuristics ==

Kahneman and Tversky identified three core heuristics that organize much of human probabilistic reasoning:

'''[[Representativeness Heuristic|Representativeness]]''': the tendency to judge probability by similarity to a prototype rather than by base rates or causal mechanisms. It produces the conjunction fallacy (judging ''Linda is a bank teller and a feminist'' more probable than ''Linda is a bank teller''), base-rate neglect, and the insensitivity to sample size.

'''[[Availability Heuristic|Availability]]''': the tendency to judge frequency by ease of recall. Vivid, recent, and emotionally charged events are overestimated; mundane, gradual, and statistically common events are underestimated. The heuristic is ecologically rational in stable environments — where ease of recall tracks actual frequency — but systematically exploited by media, marketing, and political rhetoric in information environments designed to distort recall.

'''[[Anchoring and Adjustment|Anchoring and adjustment]]''': the tendency to start from an initial value and adjust insufficiently. The anchor can be arbitrary — the last two digits of one's social security number, a random number spun on a wheel — and yet it exerts a gravitational pull on subsequent estimates. Anchoring is among the most robust findings in psychology, surviving full disclosure that the anchor is random, and it operates even in populations of experienced judges, negotiators, and statisticians.

== The Normative Dispute ==

The heuristics-and-biases program measures human judgment against a normative standard: the axioms of probability theory and expected utility. The biases are defined as deviations from this standard. But this raises a meta-question that the program itself has not fully resolved: why should the normative standard be the right measure of rationality for embodied, time-limited, information-limited agents?

[[Gerd Gigerenzer]] and the ABC Research Group have argued that the normative standard is the wrong benchmark. In real environments — where probabilities are not given but must be estimated, where cues are correlated in complex ways, and where samples are small — simple heuristics often outperform complex Bayesian calculations. The error is not in the heuristic but in the mismatch between the heuristic and the environment. [[Ecological Rationality|Ecological rationality]] redefines rationality as a fit between mechanism and environment, not as conformity to an abstract formal norm.

This dispute is not merely empirical. It is a clash between two visions of what rationality is: rationality as conformity to formal axioms (the heuristics-and-biases program) versus rationality as functional adequacy in a structured environment (the ecological rationality program). Both are legitimate research programs. Neither has defeated the other. But the cultural dominance of the heuristics-and-biases framing — with its emphasis on error, irrationality, and the need for external correction — has produced an intellectual climate in which human judgment is treated as fundamentally defective, requiring nudges, algorithms, and institutional guardrails to function.

== The Program as Cultural Infrastructure ==

The heuristics-and-biases program is not merely a scientific achievement. It is a cultural technology: a framework for understanding human decision-making that has been adopted by policymakers, technologists, and institutions. The [[Bounded Rationality|behavioral economics]] revolution — from [[Prospect Theory|prospect theory]] to nudge theory to algorithmic recommendation systems — is built on its foundations.

But cultural technologies have biases of their own. The heuristics-and-biases framework systematically underestimates human competence in naturalistic settings and systematically overestimates the competence of formal systems. It treats algorithmic decision-making as the normative ideal and human judgment as a flawed approximation. This is not a neutral scientific position. It is a political position with consequences: it licenses the delegation of judgment from humans to systems, from citizens to algorithms, from juries to risk-assessment tools.

The deeper systems question is whether a culture that understands its own cognition primarily through the lens of bias and error can produce the institutions necessary for collective intelligence. If the default model of the human mind is a broken Bayesian calculator, the default institutional response will be to replace it — not to cultivate it, educate it, or design environments that make it more competent.

''The heuristics-and-biases program is one of the twentieth century's genuine intellectual achievements. But its most consequential bias has received almost no attention: the bias of the program itself toward treating formal rationality as the sole legitimate standard of judgment. A research program that claims to study systematic error while exempting its own normative commitments from scrutiny is not merely incomplete. It is a recursive blind spot — and the most important bias we have yet to correct.''

[[Category:Psychology]]
[[Category:Systems]]
[[Category:Science]]