A Behavioral Model of Rational Choice
The sources provided offer a deep dive into Herbert Simon's seminal work, "A Behavioral Model of Rational Choice."This paper is a foundational critique of traditional economic theory, proposing a more realistic model of how humans actually make decisions. To address your request for a comprehensive exploration of these concepts (approximately 10,000 characters in depth), we must examine the shift from "economic man" to "administrative man," the mechanics of simplification, and the dynamic nature of aspiration levels.
The Critique of "Economic Man"
Traditional economic theory is built upon the postulate of the "economic man," an individual who is assumed to be both "economic" and "rational". This idealized figure possesses a "well-organized and stable system of preferences" and an "impressively clear and voluminous" knowledge of his environment. Most importantly, he has a computational skill that allows him to calculate which alternative course of action will lead him to the highest possible point on his preference scale.
However, the sources highlight growing doubts about whether this model provides a suitable foundation for understanding how firms or individuals actually behave. The "global rationality" of the economic man is increasingly seen as incompatible with the actual access to information and computational capacities possessed by real organisms. Simon argues that the state of information should be seen as a characteristic of the decision-maker himself, rather than just a feature of the environment. To stay within their computing capacity, humans must deliberately introduce simplifications into their models of reality.
Internal vs. External Constraints
In traditional models, constraints are usually viewed as "external"—things like the set of available alternatives or the relationship between an action and its pay-off. Simon suggests that we must also account for "internal" constraints, which lie "within the skin of the biological organism". These include physiological and psychological limitations, such as the maximum speed an organism can move or its limited ability to predict the future.
Because of these limits, actual human striving for rationality is, at best, a "crude and simplified approximation" of the global rationality found in complex mathematical or game-theoretical models. By examining the schemes of approximation that humans actually employ, we can learn more about the mechanisms of choice than by studying an idealized version of a perfect calculator.
The Mechanics of Rational Choice
The sources define several "primitive terms" necessary for any model of rational behavior, whether global or limited:
- A set of behavior alternatives (A): The total range of choices objectively available.
- The considered subset (A°): The limited set of alternatives the organism actually perceives or thinks about.
- Outcomes (S): The possible future states resulting from a choice.
- A pay-off function (V): The "value" or "utility" assigned to each outcome.
- Information mapping: The knowledge of which outcome will actually occur if a specific alternative is chosen.
"Classical" concepts of rationality, such as the Max-min Rule (choosing the best among the worst-case scenarios) or the Probabilistic Rule (maximizing expected value), make "severe demands" on the organism. These rules require the decision-maker to attach definite pay-offs to every possible outcome, leaving no room for "unanticipated consequences". Simon posits that there is a complete lack of evidence that humans can, or do, perform these complex computations in actual choice situations.
Simplification Through "Simple" Pay-off Functions
To bring decision-making within the range of human capability, Simon introduces the concept of "simple" pay-off functions. Instead of a complex cardinal utility scale, outcomes are categorized as (1, 0), representing "satisfactory" or "unsatisfactory," or (1, 0, -1), representing "win, draw, or lose".
A prime example is the process of selling a house. A seller might decide that $15,000 is an "acceptable" price. In this model, any offer over that amount is "satisfactory," and anything less is "unsatisfactory". This boundary is known in psychological theory as the "aspiration level". This simplification drastically reduces the computational burden because the seller no longer needs to find the highest possible offer; they only need to find the first offer that meets their criteria.
Information Gathering and Search
In reality, the mapping of alternatives to outcomes is rarely known in advance. A decision-maker must often engage in an information-gathering process to refine their understanding of the situation. If this process is costly, the decision-maker must decide how much to refine the mapping before choosing.
Simon uses chess to illustrate this. A player in the middle of a game does not seek the "best" move by calculating every possible outcome—which could involve a "septilion" variations. Instead, they search for a "good" move. They identify a subset of positions that are "clearly won" and then explore moves that might lead to those positions. By following this "good enough" strategy, a player can reduce the number of lines of play they need to examine to fewer than 100 variations—a "spectacular simplification" of the problem.
Aspiration Levels and Dynamic Adjustment
One of the most innovative aspects of Simon’s model is the dynamic adjustment of aspiration levels. Aspiration levels are not static; they change based on the history of the system. If an individual finds it easy to discover satisfactory alternatives, their aspiration level rises. Conversely, if satisfactory alternatives are difficult to find, the aspiration level falls.
This adjustment process helps guarantee that a solution will eventually be found. If no satisfactory solution is discovered at first, the resulting drop in the aspiration level eventually "brings satisfactory solutions into existence". Organisms may also adjust by broadening or narrowing the set of alternatives they consider ($A°$). A more "persistent" organism might broaden its search for new alternatives rather than lowering its standards.
Application to Organizational Theory
Simon’s "approximate" rationality provides the materials for a theory of behavior in an organizational context. There is an apparent paradox in traditional theory: if humans were globally rational, the problems of the internal structure of a firm would largely disappear, as everyone would perfectly calculate the optimal path.
The paradox vanishes when we substitute the "economic man" with the "administrative man"—a choosing organism of limited knowledge and ability. Because the administrative man must simplify the real world, discrepancies arise between his model and reality. These discrepancies serve to explain the necessity of organizational structure and the phenomena of organizational behavior. Organizations exist, in part, to provide the environment and simplifications that allow limited human organisms to make "intendedly" rational decisions.
Normative and Descriptive Value
The model is not just descriptive (explaining how things are) but also has normative value (suggesting how things should be). By understanding the limits of human rationality, we can better design computing equipment and administrative processes.
Simon notes that comparing the intelligence of a computer to a human is difficult. On factors where a computer might score as a "genius," a human might appear as a "moron," and vice versa. Understanding different definitions of rationality can suggest new directions for the design of computers that excel in the areas where they are currently "moronic".
Mathematical Appendix: Rationality at Various Levels
The appendix of the paper explores how an individual might rationally determine an "acceptable" pay-off. Using the house-selling example, Simon demonstrates that if a seller has information about the probability distribution of offers, they can mathematically set a daily "acceptance price" ($d(k)$) that maximizes the expected sales price.
However, even this "rational" setting of a price involves a high level of information—the seller needs to know the probability distribution of offers for all future time periods. A "bumbling" kind of rationality is more common: the seller sets a high price, watches the offers that come in, and gradually adjusts the price up or down without ever making a formal probability calculation. This is the kind of adjustment that humans find "good enough" and are capable of exercising in a wide range of practical circumstances.