Bounded Rationality

Conceptual Overview

Bounded rationality has been particularly important to the foundational works, as well as later developments, in organization theory. It is also a crucial element to an understanding of decisions in organizations that often does not exist in neoclassical theory because of its unrealistic assumptions. Other developments in economics—in game theory and macroeconomics, for example—have developed the idea of BR in a more neoclassical direction, but they are not about decision making in organizations, and so shall not be discussed here.

Notwithstanding earlier discussions of the limitations of rationality, bounded rationality was introduced by Simon in 1955 in his critique of the assumption in [Page 107]economics of perfect information and unlimited computational capability. He wanted to replace the assumption of global rationality with one that better corresponded with how humans (and other choosing organisms) made decisions, their computational limitations, and how they accessed information in their current environments. In Simon's illustration of the problem of global rationality, the influence of his 1947 ideas (outlined in Administrative Behavior) is clear: Decisions are reasoned and intended to be rational, yet they are limited. He first suggests a simple and very general model of behavioral choice that analyzes choosing organisms (such as humans) in terms of basic properties to understand what is meant by rational behavior. He introduces simplifying assumptions, such as choice alternatives, the payoff function, possible future states, and the subset of choice alternatives that is considered, as well as information about the probability that a particular outcome will lead to a particular choice. But immediately afterward, he turns to the simplifications of this model, stressing that, upon careful examination, the model makes severe demands on the capacity of the agency or organism exercising choice. In rational choice models, definite payoffs (or at least a definite range of payoffs) need to be attached to each possible outcome, but Simon suggests that there is no evidence that such computations exist in complex situations of human choice. Because of the lack of computational capacity, decision makers have to simplify the structure of their decisions (and thus satisfice)—one of the most important lessons of BR.

...