heuristics a rule or solution adopted to reduce the complexity of computational tasks, thereby reducing demands on resources such as time, memory, and attention. If an algorithm is a procedure yielding a correct solution to a problem, then a heuristic procedure may not reach a solution even if there is one, or may provide an incorrect answer. The reliability of heuristics varies between domains; the resulting biases are predictable, and provide information about system design. Chess, for example, is a finite game with a finite number of possible positions, but there is no known algorithm for finding the optimal move. Computers and humans both employ heuristics in evaluating intermediate moves, relying on a few significant cues to game quality, such as safety of the king, material balance, and center control. The use of these criteria simplifies the problem, making it computationally tractable. They are heuristic guides, reliable but limited in success. There is no guarantee that the result will be the best move or even good. They are nonetheless satisfactory for competent chess. Work on human judgment indicates a similar moral. Examples of judgmental infelicities support the view that human reasoning systematically violates standards for statistical reasoning, ignoring base rates, sample size, and correlations. Experimental results suggest that humans utilize judgmental heuristics in gauging probabilities, such as representativeness, or the degree to which an individual or event resembles a prototypical member of a category. Such heuristics produce reasonable judgments in many cases, but are of limited validity when measured by a Bayesian standard. Judgmental heuristics are biased and subject to systemic errors. Experimental support for the importance of these heuristics depends on cases in which subjects deviate from the normative standard. See also BAYESIAN RATIONALITY, EMPIRI- CAL DECISION THEORY. R.C.R.