1 Judgment and Decision-Making Under Uncertainty

Judgment, decisions, risk, and uncertainty

People often need to estimate what may happen and choose what to do without knowing the outcome. A judgment estimates something, such as the chance of rain, while a decision selects an action using beliefs about possible outcomes and their value.

In risk, probabilities are known or reasonably estimable. In uncertainty, probabilities may be unknown or poorly specified. Everyday choices can involve both.

Heuristics and biases

Heuristics simplify judgment and decision-making when time, information, or effort is limited. They can be efficient and useful, and they are not errors by definition. A bias is a systematic tendency that can skew judgment, sometimes when a heuristic is applied in a situation where it gives a misleading answer.

Representativeness and base rates

Representativeness estimates likelihood by how much something resembles a category or stereotype. This can lead to neglecting base rates, or how common the category is. For example, a quiet, book-loving person may seem more likely to be a librarian than a salesperson, but resemblance alone does not establish which occupation is more probable; the prevalence of each occupation also matters.

Availability

Availability estimates frequency or likelihood according to how easily examples come to mind. Recent or vivid events may feel more common than less memorable events, even when they are not.

Anchoring and adjustment

Anchoring and adjustment begins with an initial number or estimate and adjusts from it. If the adjustment is insufficient, the initial anchor can pull the final estimate toward itself, even when the anchor is arbitrary.

These are tendencies, not inevitable mistakes. A vivid example may sometimes be genuinely informative, and a familiar category may sometimes provide useful evidence. Good judgment depends on checking whether a cue is relevant and whether important information, especially base rates, has been overlooked.

Expected utility as a benchmark

A normative model specifies how a choice should be made under stated assumptions. Expected utility theory is a common benchmark: identify possible outcomes, assign probabilities and utilities (values) to them, and choose the option with the greatest probability-weighted utility.

Here, $EU(a)$ is the expected utility of action $a$, $p_i$ is the probability of outcome $o_i$ given that action, and $U(o_i)$ is that outcome’s utility.

For example, when comparing two travel routes, a decision-maker could consider each route’s possible travel times, their probabilities, and the value placed on arriving quickly versus avoiding delay. The preferred route need not have the shortest average time if reliability or other outcomes matter more.

Expected utility is a benchmark, not a claim that people always calculate probabilities and values explicitly. It cannot supply missing facts or determine a person’s values. If probabilities are unknown, or if the model leaves out outcomes that matter, its recommendation may be inadequate.

A descriptive account of choice

Prospect theory was developed as a descriptive alternative for choices under risk. It accounts for observed patterns such as sensitivity to gains and losses and the disproportionate influence of some probabilities. Unlike a normative benchmark, a descriptive model helps explain how people actually choose.

Using the concepts together

Heuristics describe simplifying strategies people may use; biases describe systematic distortions that can result; normative models provide standards for evaluating choices. These concepts answer different questions and should not be treated as interchangeable.

To improve a decision, make the options and outcomes explicit, check relevant base rates and evidence, state what is uncertain, and clarify which outcomes matter. A heuristic can be a practical starting point; comparing the resulting judgment with a suitable normative model can show where more careful reasoning is needed.