Bounded rationality is an approach to decision theory that explains choice in terms of the knowledge and computational capacities actually available to decision-makers. Developed by Herbert A. Simon, it examines how people pursue goals when they cannot identify every alternative, predict every consequence, or calculate the best possible choice. It is central to behavioral economics and connects economic analysis with psychology. (web.stanford.edu)
Origins
Simon developed the approach through research on administrative and organizational decision-making. His Administrative Behavior (1947) treated organizations as mechanisms for coping with complexity and uncertainty. His 1955 paper “A Behavioral Model of Rational Choice” advanced an alternative to globally rational choice, and he introduced the term “bounded rationality” in 1957. In 1978, he received the Nobel Memorial Prize in Economic Sciences for research into decision-making within economic organizations. (nobelprize.org)
Departure from idealized rational choice
A standard subjective expected-utility model assumes a specified set of alternatives, known subjective probability distributions over their consequences, and selection of the alternative with the highest expected value of utility. Bounded-rationality models relax one or more of these assumptions: alternatives may need to be discovered, consequences may be difficult to estimate, and comprehensive comparison may exceed cognitive capacity. (web.stanford.edu)
The distinction is not simply between having complete and incomplete information. Even with extensive information, evaluating alternatives can be computationally infeasible. Searching for information and deciding which calculations to perform also consume resources. Bounded rationality therefore concerns the feasibility of the decision process, not merely the information available at its outset. (iiif.library.cmu.edu)
Satisficing and heuristics
A characteristic strategy is satisficing: searching until an alternative meets an acceptable standard, rather than continuing until the best alternative has been established. The standard is an aspiration level, which supplies a stopping rule. For decisions involving several goals, an acceptable option may need to satisfy a threshold on each dimension. (iiif.library.cmu.edu)
For example, a hypothetical housing search might require a dwelling to meet limits on price and commuting time while providing a minimum amount of space. Accepting the first dwelling that satisfies these requirements is satisficing; identifying the highest-ranked dwelling among every possible option is optimization.
Aspiration levels can change with experience: easy success may raise them, while persistent failure may lower them. Search can also employ a heuristic, or selective rule of thumb, to direct attention toward promising alternatives. Such rules reduce computational demands without necessarily guaranteeing an optimal—or even satisfactory—result. (iiif.library.cmu.edu)
Satisficing is one model of bounded rationality, not its complete definition. Other approaches examine simplified judgment rules, limited reasoning, or the balance between decision quality and computational effort. (plato.stanford.edu)
Procedural rationality and organizations
Simon distinguished substantive rationality, which evaluates the choice relative to goals and circumstances, from procedural rationality, which examines how the choice is produced. A procedural account seeks to explain the actual search, evaluation, and selection processes rather than only predicting their final outcome. (web.stanford.edu)
Organizations can address individual limitations by dividing decision-making among specialists, establishing communication and authority structures, and replacing broad objectives with manageable subgoals. These arrangements help coordinate activity that exceeds any one person’s capacity to comprehend and calculate. (nobelprize.org)
In the theory of the firm, this approach permits models in which decision-makers seek acceptable solutions to several, sometimes conflicting, organizational problems rather than calculating a single globally profit-maximizing action. (nobelprize.org)
Interpretation and empirical assessment
One interpretation treats bounded rationality as optimization under information and computational constraints. Another emphasizes explicit procedures that are feasible without calculating an optimal allocation of reasoning effort. The latter observes that finding the optimal search strategy can itself require unavailable knowledge and computation. (plato.stanford.edu)
Bounded rationality is therefore a family of approaches rather than a single predictive model. Particular models require specifications of search procedures, aspiration levels, and adjustment mechanisms. Their explanatory value depends on whether those specifications accurately represent observed decision-making. (iiif.library.cmu.edu)
References
- Bounded Rationality — Herbert A. Simonweb.stanford.edu
- Bounded Rationality — Stanford Encyclopedia of Philosophyplato.stanford.edu
- Herbert A. Simon — Prize Lecturenobelprize.org
- The Prize in Economics 1978 — Press releasenobelprize.org
- Satisficing — Herbert A. Simoniiif.library.cmu.edu