A formal method that represents a single decision‑maker's choice under uncertainty by specifying actions, states of the world, outcomes, the decision‑maker's beliefs (probabilities) and preferences (utility or loss function), together with a decision rule (e.g., maximize expected utility, minimize expected loss) to evaluate and compare actions.

Principle

Given well‑specified beliefs and a preference representation, a decision rule (most commonly expected utility maximization) orders actions by their expected value computed from the utility function and probability distribution; value of information and dominance relations follow directly from this representation.

Demonstration

Illustrative scenario → A clinician must choose Treatment A or B for a patient with uncertain diagnosis. Specify actions = {A,B}; states = {Disease X, not X}; assign probabilities to states and utilities to outcomes (benefit/harm). Compute expected utility for each treatment and select the action with higher expected utility. Consequence: makes explicit the role of prior probability and outcome valuation in the treatment choice and quantifies the value of diagnostic tests.

Misapplication

Applying a chosen decision rule (e.g., expected utility) without validating that preferences are representable by the required axioms, or using poorly elicited utilities/probabilities; the error is treating the formal output as authoritative despite misspecified beliefs or state‑dependent utilities.

Consequence

Clarifies tradeoffs between outcomes, allows computation of value of information and optimal policies under given assumptions, and provides a normative benchmark for coherence; however, prescriptive recommendations are sensitive to how utilities and probabilities are elicited and to ambiguity about states.

Reversal

In contexts of ambiguity (unknown probabilities), interdependent preferences, or when preferences violate expected‑utility axioms (e.g., independence), alternative models (ambiguity‑sensitive criteria, non‑expected utility frameworks) or multi‑criterion decision rules may be more appropriate.

Boundary

Clearly within: single‑agent choices under quantified uncertainty where beliefs and preferences can be modeled. Boundary case: decisions embedded in strategic environments where others' choices matter (requires game‑theoretic extension). Clearly outside: descriptive accounts of heuristics and biases when the goal is to model actual cognitive processes rather than normative prescription.

Semantic Tension

Normative coherence (axiomatic rationality and internal consistency) ↔ descriptive realism (empirical deviations from axioms, ambiguity attitudes); the formal ideal may conflict with observed decision behavior.

Synthesis

Decision‑theoretic analysis supplies a compact formal language to expose and compare the roles of beliefs and values in choice, making both prescriptive recommendations and diagnostic tests possible, but its applicability depends on the fidelity of elicited probabilities and utilities.