Expectation Principle

The Expectation Principle is the rule that evaluates an uncertain prospect by taking the value of each possible outcome and weighting it by the outcome’s probability. In its simplest form, the expected value of a gamble is the average of its monetary outcomes, each multiplied by the chance of receiving it. The principle became a foundation of modern decision theory because it promises a consistent way to compare uncertain options. Yet the evidence presented in the book shows that people do not experience probability as a neutral measuring scale: they overweight possibilities, underweight near-certainties, and allow framing and reference points to alter choices between objectively equivalent prospects.

Definition and Mathematical Form

The principle emerged from the older practice of assessing gambles by their expected monetary value. A prospect offering a 20% chance to win $1,000 and a 75% chance to win $100 has an expected value of $275, obtained by multiplying each outcome by its probability and adding the products. Daniel Bernoulli retained this method of assigning weights to outcomes but applied it to psychological value rather than money itself. In his account, the utility of a gamble is the average of the utilities of its possible outcomes, with each utility weighted by the relevant probability.

Bernoulli’s theory also supplied an explanation for risk aversion. In the standard example, a person chooses between an 85% chance of winning $1,000, with a 15% chance of winning nothing, and $800 for certain. The gamble has a higher monetary expectation, $850 compared with $800, but most people prefer the sure gain. Bernoulli argued that subjective value is a concave function of money: the difference between $100 and $200 matters more than the difference between $1,100 and $1,200. Because of this diminishing sensitivity to wealth, the subjective value of receiving $800 for certain can exceed the weighted subjective value of the gamble.

The same logic predicts risk seeking in losses. When people must choose between an 85% chance of losing $1,000, with a 15% chance of losing nothing, and a sure loss of $800, most prefer the gamble even though its expected loss is $850, worse than the sure loss of $800. The contrast between risk aversion for gains and risk seeking for losses became an important feature of later work on decision making. It also showed that a rule that is mathematically simple can generate psychologically different responses depending on whether the outcomes are experienced as gains or losses.

The Normative Rule and Its Authority

The Expectation Principle acquired stronger authority in the axiomatic theory of rational choice developed by John von Neumann and Oskar Morgenstern. Their theory treated preferences as subject to principles such as transitivity and substitution, alongside dominance and invariance. The axiomatic argument held that weighting uncertain outcomes in a way that is not strictly proportional to probability produces inconsistency. Expected utility theory therefore placed the principle at the center of the rational-agent model used in economics and other social sciences.

Its appeal is not merely computational. If two prospects have the same outcomes and probabilities, the principle promises a stable basis for comparison, independent of the wording used to describe them. It belongs to a broader ideal of coherence in which preferences should remain logically compatible across equivalent representations. This ideal became a benchmark against which actual human choices could be measured, even when the choices themselves repeatedly violated it.

The book presents the principle as both indispensable and psychologically inadequate. A person who wishes to make a formally rational decision is expected to use probabilities as weights, but ordinary decision makers do not reliably do so. This does not mean that every departure from the principle is foolish or that people are simply irrational. It means that the human mind applies a different psychology to uncertain outcomes than the rational-agent model assumes.

Psychological Violations of Probability Weighting

The clearest challenge concerns changes in the probability of winning $1 million. The Expectation Principle says that a five-percentage-point increase should have the same effect whether the chance rises from 0% to 5%, from 5% to 10%, from 60% to 65%, or from 95% to 100%. Human judgments do not follow that pattern. Most people regard the move from 0% to 5% and the move from 95% to 100% as far more significant than either of the intermediate changes.

The move from 0% to 5% creates a possibility where none existed before. This possibility effect gives a very small chance of a large outcome disproportionate psychological weight. Lottery purchases illustrate the pattern: people may pay substantially more for a tiny chance of winning a large prize than the mathematical expectation of that chance would justify. The move from 5% to 10% doubles the probability, but people do not generally experience the prospect as twice as valuable. The psychological transformation involved in making an outcome possible is larger than the corresponding numerical change.

The move from 95% to 100% produces the certainty effect. Near-certainty is treated as less valuable than actual certainty, so people may pay heavily to convert an almost guaranteed outcome into a guaranteed one. The book illustrates this with a $1 million inheritance contested by a stepsister. A lawyer estimates a 95% chance of winning, implying an expected value of $950,000, but a company offers $910,000 immediately for the case. The discount is $40,000 below the expected value, yet the offer can appear attractive because it removes the remaining uncertainty. The existence of a structured-settlement industry reflects the willingness to pay a substantial price for certainty.

The same pattern governs losses. A 5% risk that a loved one will require an amputation during surgery can feel much worse than half as bad as a 10% risk, because the small risk has become a vivid possibility. At the other end, the difference between a 95% risk of disaster and certain disaster preserves a psychologically important sliver of hope. Thus, overweighting small probabilities increases the appeal both of gambling and of insurance: the same sensitivity to unlikely outcomes can make a remote prize seem worth pursuing and a remote catastrophe seem worth paying to eliminate.

The resulting conclusion is direct: decision weights are not identical to probabilities. Improbable outcomes are overweighted, while outcomes that are almost certain are underweighted relative to certainty itself. The Expectation Principle remains a precise rule for probability-weighted value, but it is poor as a description of how people actually evaluate risk.

Reference Points, Framing, and Invariance

The principle also encounters difficulty when outcomes are described relative to a reference point. Daniel Kahneman and Amos Tversky’s prospect theory departed from utility theory by adding the earlier state against which gains and losses are evaluated. A person’s wealth alone is insufficient to determine the value of an outcome; the relevant comparison may be the status quo, an expectation, or an outcome to which the person feels entitled. The same final wealth can therefore be experienced as a gain in one setting and a loss in another.

This reference-point structure helps explain why the same objective choice can produce different preferences under different descriptions. In the Asian disease problem, one version describes a program as saving 200 people for certain or offering a one-third chance of saving 600. Respondents predominantly choose the sure gain. An equivalent version describes the alternatives as 400 people dying for certain or a one-third chance that nobody dies and a two-thirds chance that 600 die. Respondents then predominantly choose the gamble, showing risk seeking in the domain of losses. The material outcomes are equivalent, but the reference state and wording shift the psychological meaning from gains to losses.

This violates invariance, the requirement that preferences between equivalent prospects should not depend on how those prospects are described. The violation is pervasive among both sophisticated and naive respondents and persists even when the same people answer both versions within minutes. When confronted with their contradictory answers, respondents commonly want all three things at once: risk aversion in the lives-saved version, risk seeking in the lives-lost version, and consistency between the two. The conflict resembles a perceptual illusion more than a simple arithmetic mistake.

The book also shows how framing can produce violations of dominance. In paired decisions involving sure gains and sure losses, 73% of respondents chose the combination of options A and D, while only 3% chose B and C. Yet when the two decisions are combined, the supposedly preferred conjunction is dominated by the rejected one. Adding the sure gain to one gamble produces the very prospect rejected in the earlier problem, while adding the sure loss to the other produces the alternative that had been preferred. The failure arises because people evaluate decisions separately and respond to their gain-and-loss framing rather than integrating their outcomes into a single actuarial comparison.

Limits, Uses, and Broader Significance

The book considers two possible ways of protecting the ideal of invariance. One is to convert every equivalent version of a problem into a canonical representation, such as total assets rather than gains and losses. That recommendation is easier to state than to follow. Intuitive decision makers naturally think in terms of gains and losses, and combining all outcomes across concurrent decisions exceeds the capabilities of intuitive computation even in simple cases. The difficulty becomes greater in questions of safety, health, and quality of life, where there may be no obvious single representation of the consequences.

The other possibility is to evaluate choices by their actuarial consequences rather than their psychological consequences. That approach has some appeal when the outcomes are human lives, but it is inadequate for financial decisions and impossible when outcomes lack an objective metric. The conclusion is therefore not that the Expectation Principle should simply be discarded. Rather, confidence in a choice cannot guarantee that the same choice will survive a change of frame. A practical safeguard is to restate a decision deliberately in more than one way and test whether the preference remains stable.

The principle’s continuing importance also appears in the study of experts and institutions. Philip Tetlock’s twenty-year investigation collected more than 80,000 probability judgments from 284 people who earned their living commenting on political and economic trends. These experts performed worse than they would have if they had distributed equal probabilities across three possible outcomes: continuation of the status quo, more of the relevant condition, or less of it. Even in their areas of specialization, they were not significantly better than nonspecialists, and greater knowledge could intensify the illusion of predictive skill.

A related survey gathered 11,600 forecasts from chief financial officers of large corporations about the following year’s Standard and Poor’s index. The correlation between their estimates and the actual market value was slightly below zero. Their 80% confidence intervals were also badly miscalibrated: 67% of outcomes fell outside the interval, rather than the roughly 20% expected from an honestly calibrated interval. These findings connect probability weighting with overconfidence. When information that comes to mind is used to construct a coherent story, information that never entered awareness cannot easily be included in the estimate.

The Expectation Principle thus occupies a productive boundary between normative theory and descriptive psychology. It defines how probabilities should contribute to a coherent evaluation of risk, while the evidence shows why real people depart from that standard. Prospect theory, developed through the work of Daniel Kahneman and Amos Tversky, did not merely reject the principle; it retained the problem of weighting outcomes while identifying reference points, gains and losses, framing, possibility, and certainty as forces that reshape judgment. The result is a sharper understanding of rational choice: the principle remains a formal ideal, but any serious account of human decisions must explain why that ideal is so difficult to enact.