Rationality
From DDL Wiki
In the study of decision-making, rationality is a characteristic of a decision-making process. There are different types of rationality that correspond to the following definition (Stirling, 2003): A rational decision is one that conforms either to a set of general principles that govern preferences or to a set of rules that govern behavior.
Substantive rationality is the paradigm that guides formal decision analysis. It deals with principles about preferences. First, for all of the possible options, the decision-maker has a total ordering over them. Second, the decision-maker should choose the option that is most preferred.
Procedural rationality deals with decision-making processes. Procedural rationality implies that a decision-maker uses specific rules or procedures to make a choice. This is context specific, because the rules or procedures that make sense in one domain may be poor choices in another.
Bounded rationality starts with the observation that information and computational power (be it computers or people) are limited in the real-world, and this prevents complete optimization. In this paradigm, satisficing is seen as an appropriate strategy. Many of the choice strategies reflect this paradigm.
Intrinsic rationality looks at each alternative by itself and considers whether the expected benefits of the alternative exceed the expected losses. If the alternative is a net gain, then keep it, else discard it. Intrinsic rationality allows a decision-maker to create a set of (intrinsically) rational solutions (instead of just one optimal or satisfactory solution).
Gigerenzer and Todd (1999) also discuss different models of rationality and present a different classification scheme. They first distinguish between Demons and Bounded Rationality. The models called demons (after supernatural, omniscient beings) are those that assume that human beings have unlimited power to find and evaluate alternatives (especially those that seem to have uncertain outcomes). This class they separate into two types: unbounded rationality and optimization under constraints.
Unbounded rationality assumes that humans have unlimited time, knowledge, and computational capacity and do not consider the cost of searching.
Optimization under constraints assumes that the decision-maker considers the costs and benefits of searching for information but otherwise still has unlimited knowledge and computational capacity.
The models in bounded rationality include the two overlapping categories of satisficing and fast and frugal heuristics.
Satisficing is one type of bounded rationality. In this model, the decision-maker determines minimum requirements on one or more attributes and searches for a solution until he finds one that meets all of the minimum requirements. The decision-maker stops with the first satisfactory one.
Fast and frugal heuristics (some of which may be satisficing) employ a minimum of time, knowledge, and computation to make adaptive choices in real environments. Moreover, they limit their search of objects or information using easily-computable stopping rules, and they make their choices with easily-computable decision rules.
References
Stirling, Wynn C., Satisficing Games and Decision Making, Cambridge University Press, Cambridge, 2003.
Gigerenzer, Gerd, Peter M. Todd, and the ABC Research Group, Simple Heuristics That Make Us Smart, Oxford University Press, 1999. A precis is available online at http://www-abc.mpib-berlin.mpg.de/users/ptodd/SimpleHeuristics.BBS/
