Decision theory
|
Decision theory | |
|---|---|
|
| |
| Type | Normative and analytical framework |
| Field | Philosophy; Economics; Mathematics |
| Core idea | Formal analysis of how choices should be made given goals, preferences, and uncertainty |
| Assumptions | Agents have preferences; options can be compared using decision criteria |
| Status | Established framework |
| Related | Rationality; Probability theory; Utility; Game theory |
Decision theory is a framework for analyzing and evaluating choices under conditions of uncertainty, risk, or incomplete information. It provides formal tools for comparing options based on their expected outcomes relative to an agent’s goals, preferences, and beliefs.
Decision theory is used in philosophy, economics, statistics, and artificial intelligence to clarify what it means to choose rationally and how decisions can be justified.
Core idea
At its core, decision theory addresses how an agent ought to choose among available options. Each option is evaluated in terms of possible outcomes, the likelihood of those outcomes, and their desirability to the agent.
Decision theory does not prescribe what goals agents should have; it analyzes how choices should be made given those goals.
Preferences and utility
Decision theory typically represents an agent’s preferences using a utility function, which assigns numerical values to outcomes reflecting their relative desirability.
Utility is a formal representation of preference, not a measure of pleasure or moral value, and it allows different options to be compared systematically.
Beliefs and probability
Uncertainty about outcomes is represented using probabilities. These probabilities encode an agent’s beliefs about how likely different outcomes are, given available information.
Decision theory relies on probability theory to combine beliefs and preferences into evaluative criteria.
Expected utility
A central concept in decision theory is expected utility, which combines the utility of outcomes with their probabilities. According to expected utility theory, rational agents should choose the option with the highest expected utility.
This principle provides a normative standard for rational choice under uncertainty.
Normative and descriptive perspectives
Decision theory can be understood in both normative and descriptive terms. Normative decision theory specifies how agents ought to choose, while descriptive approaches study how agents actually choose.
Empirical findings often reveal systematic deviations from normative models, motivating alternative frameworks.
Decision theory and rationality
Decision theory is closely related to rationality, particularly practical rationality. It provides formal criteria for evaluating whether choices are consistent, coherent, and responsive to reasons.
Disagreements arise over whether decision-theoretic norms capture all relevant aspects of rational choice.
Extensions and variants
Various extensions of decision theory address limitations of basic models. These include decision-making under ambiguity, bounded rationality, and dynamic decision problems involving time and learning.
Different variants relax or modify assumptions about information, computation, or preference stability.
Decision theory and ethics
Decision-theoretic tools are sometimes applied to ethical reasoning, policy analysis, and social choice. Such applications raise questions about how values should be represented and whether utility captures morally relevant considerations.
These uses highlight the distinction between formal decision criteria and substantive ethical judgment.
Limits and critiques
Critiques of decision theory argue that its assumptions about preference completeness, consistency, and quantification may not hold in real-world contexts. Others question whether all rational considerations can be reduced to utility and probability.
These critiques motivate pluralistic approaches to rational choice.
Status
Decision theory is an established analytical framework with wide applicability. Its primary contribution is clarifying the structure of rational choice and making assumptions about belief, preference, and uncertainty explicit.