Bounded Rationality

Origins and Core Idea

The concept of bounded rationality was introduced by economist and cognitive scientist Herbert A. Simon in the mid-20th century, as a challenge to the then-dominant model of unbounded or perfect rationality in economics and decision theory. Traditional rational choice models assumed that agents have complete information, unlimited computational power, and sufficient time to evaluate all possible options and select the utility-maximizing one. Simon argued that this picture is psychologically unrealistic and empirically misleading.

According to bounded rationality, human beings are intendedly rational: they generally aim to make good decisions, but their reasoning is bounded by three main constraints:

1. Limited information: agents rarely have access to all relevant facts or probabilities.
2. Limited cognitive capacity: even given information, they cannot process or compute all implications.
3. Limited time: decisions are often made under deadlines or pressure.

Bounded rationality therefore treats decision-making as a problem of adaptation under constraints rather than perfect optimization. This idea has been influential across economics, philosophy of rationality, cognitive science, political science, and artificial intelligence.

Heuristics and Satisficing

A central feature of bounded rationality is the use of heuristics, or rule-of-thumb strategies, in place of exhaustive computation. Simon and later researchers such as Gerd Gigerenzer describe heuristics as simple decision rules that exploit structural features of environments. Examples include:

• Recognition heuristic: if one of two objects is recognized and the other is not, infer that the recognized one has the higher value on some criterion (e.g., city size).
• Take-the-best heuristic: compare options on the most important cue first; if that discriminates, stop and decide, otherwise move to the next cue.

These heuristics often require less information and computation than formal optimization procedures.

Simon introduced the notion of satisficing as an alternative to maximizing. A satisficing agent looks for an option that is “good enough” relative to an aspiration level, rather than the best possible option. This can involve:

• Setting a threshold of acceptability (e.g., a minimum salary or quality level).
• Searching options sequentially.
• Stopping the search once an option meets or exceeds the threshold.

Satisficing is an example of a sequential search strategy: given constraints, it can be more efficient overall than trying to survey and compare every option. Proponents argue that such strategies are often ecologically rational—well-adapted to the structure of real-world environments, where information is costly and uncertainty pervasive.

Normative Debates and Criticisms

Bounded rationality raises contentious questions about what counts as rational:

• Descriptive vs. normative: Some interpret bounded rationality primarily as a descriptive theory (how people in fact decide), whereas others argue it should also revise the normative standards of rationality (how people ought to decide, given their limits).
• Second-best rationality: One view holds that ideal rationality is still defined by full optimization, and bounded rationality describes unavoidable deviations from that ideal. On this reading, human decision-making is systematically suboptimal, though often approximately good.
• Constraint-sensitive norms: Another view claims that norms of rationality should be indexed to an agent’s feasible capacities and environment. What is rational for a human, given our limits, may be the best implementable strategy rather than the utility-maximizing one in principle.

Critics contend that the theory of bounded rationality is sometimes vague or under-specified, offering a general slogan (people have limits) without a unified formal framework. Others argue that too-flexible appeals to bounded rationality risk immunizing theories from falsification: any observed behavior can be labeled “rational under constraints.”

Research in behavioral economics and cognitive psychology, notably by Daniel Kahneman and Amos Tversky, has complicated the picture further. Their work on heuristics and biases suggests that many of the shortcuts people use lead to systematic errors (e.g., framing effects, loss aversion, overconfidence). While both approaches accept cognitive limits, Gigerenzer and colleagues emphasize “fast and frugal” heuristics as often adaptive, whereas the heuristics-and-biases tradition stresses departures from ideal normative standards such as Bayesian updating and expected utility maximization.

Relations to Other Theories of Rationality

Bounded rationality interacts with several other important ideas in philosophy and the social sciences:

• Classical rational choice theory: Bounded rationality developed as a modification of this framework. Some models incorporate bounds by limiting the size of information sets, adding search costs, or using simplified belief-updating rules.
• Procedural vs. substantive rationality: Simon distinguished substantive rationality (quality of outcomes relative to goals) from procedural rationality (quality of the processes of reasoning). Bounded rationality often focuses on procedures that are computationally plausible for real agents.
• Computational and algorithmic perspectives: In AI and cognitive science, bounded rationality is tied to computational complexity (what problems are tractable) and to the idea of agents using algorithms that approximate optimal solutions within resource limits.
• Ecological rationality: This approach evaluates rationality relative to specific environments. A heuristic can be rational if it performs well in the environments an agent typically faces, even if it deviates from formally optimal rules in abstract cases.
• Skepticism about idealization: Philosophers of science and economics debate whether assuming unbounded rationality is a useful idealization for building explanatory models, or whether bounded rationality should be the primary foundation of theories about human behavior.

Overall, bounded rationality reframes rationality as constrained, context-sensitive, and process-oriented, emphasizing how real agents navigate complex environments with limited cognitive and informational resources. It remains central to ongoing discussions about decision-making, normativity, and the appropriate standards by which to evaluate human reasoning.