The average psychological satisfaction a decision-maker obtains from a risky option (such as an investment or insurance). It is not the money actually earned, but the weighted average of your subjective feelings about that money.
u(x) (Utility Function)A mathematical rule that converts objective outcomes (such as money x) into subjective value. For example, the happiness brought by earning 1 million is not twice that of earning 500,000, but less (this is "diminishing marginal utility").
x_i (the i-th possible outcome)x_i The specific gain or loss that the option may bring, such as "earn 100 yuan" or "lose 50 yuan."
p_i (the probability of the i-th outcome)The objective likelihood of that outcome occurring; the sum of all p_i must equal 1.
p and 1−p (two-outcome special case)When there are only two outcomes (n=2), p is the probability of outcome x_1 occurring, and 1−p is the probability of outcome x_2 occurring.
Expected Utility Theory is the cornerstone of decision theory. It tells us: when facing uncertain outcomes, a rational decision-maker does not directly compare expected monetary value, but rather compares expected utility value. The utility function u(x) characterizes the subjective satisfaction with wealth, and a concave utility function implies risk aversion—the pain of a loss is greater than the pleasure of an equal gain.
Utility Function Settings
Decision Scenario Parameters
Real-Time Calculation Results
About Expected Utility Theory
Expected Utility Theory was formally proposed by John von Neumann and Oskar Morgenstern in 1944 in Theory of Games and Economic Behavior, and is a core pillar of microeconomics and behavioral decision theory. It provides a mathematical framework for understanding how people make rational decisions under uncertainty.
Plain-Language Introduction
Imagine you face a choice: receive 100 yuan for sure, or flip a coin—heads gets 200 yuan, tails gets 0 yuan. The "expected value" of both is 100 yuan, but most people would choose the certain 100 yuan. Why? Because the satisfaction brought by money (utility) does not grow linearly— the increase in happiness from 0 to 100 yuan is far greater than from 100 to 200 yuan. Expected utility theory precisely describes this psychology with mathematics: we use a concave utility function to characterize the phenomenon that "the extra happiness brought by each additional unit of money becomes smaller and smaller," thereby explaining why people are risk averse.
What Problem Does It Solve?
Before expected utility theory appeared, people used "expected value" to make decisions, but this could not explain the St. Petersburg Paradox (people are unwilling to pay a high price for a game with infinite expected value). By introducing a subjective utility function, expected utility theory successfully explained:
- Why people buy insurance (risk aversion)
- Why the same amount of money has different "value" to different people
- How to make rational trade-offs between risk and return
Why Is It Important?
Expected utility theory is the theoretical foundation of modern economics and finance. It is the basis of asset pricing models (such as CAPM), insurance actuarial science, portfolio theory, auction design, and mechanism design. In artificial intelligence and reinforcement learning, the expected utility framework is also widely used to construct the objective functions of agents. Understanding it is equivalent to obtaining the key to understanding "rational decision-making."
Application Scenarios
- Investment decisions: Allocate assets among stocks, bonds, and cash
- Insurance pricing: Calculate the upper limit of the premium that the insured is willing to pay
- Medical decisions: Weigh the efficacy of treatment options against the risk of side effects
- Public policy: Evaluate the cost-benefit of environmental regulation and safety standards
- Game theory: Analyze competitors' strategies in uncertain environments
Limitations and Deficiencies
- Allais Paradox: Experiments show that people systematically deviate from the predictions of expected utility theory in certain situations
- Framing Effect: Describing the same problem in different ways changes choices, which the theory cannot explain
- Loss Aversion: People are far more sensitive to losses than to gains (prospect theory improved on this)
- Ambiguity Aversion: When probabilities are unknown (ambiguity), expected utility theory is difficult to apply directly
- Overly strong assumptions: It requires decision-makers to have complete information and always be rational, which is difficult to satisfy in reality