Framing effect
The framing effect is a cognitive bias in which logically equivalent descriptions of the same decision problem produce systematically different choices. A presentation emphasizing gains commonly elicits greater preference for a certain outcome, whereas an equivalent presentation emphasizing losses commonly increases preference for a risky outcome. The effect demonstrates that human judgment depends not only on final states but also on the reference point and linguistic frame through which those states are represented.
Framing effects occur when alternative descriptions preserve the decision’s relevant quantitative structure. A choice between saving 200 of 600 people and accepting a one-third probability that all 600 will be saved is mathematically equivalent to a choice between allowing 400 people to die and accepting a two-thirds probability that all 600 will die. Differences in responses to these formulations therefore cannot be attributed to differences in stated probabilities or final outcomes. They arise from the classification of outcomes as gains or losses.
Theoretical basis
The principal account of the framing effect derives from prospect theory, developed by Daniel Kahneman and Amos Tversky as an alternative to descriptive interpretations of expected utility theory. Prospect theory represents outcomes as changes relative to a reference point rather than as absolute levels of wealth or welfare.
A simplified prospect-theoretic value function is written as
[ v(x)= \begin{cases} x^\alpha, & x \geq 0,\ -\lambda(-x)^\beta, & x<0, \end{cases} ]
where (x) denotes the outcome relative to the reference point. The exponents (\alpha) and (\beta) describe diminishing sensitivity as gains or losses increase in magnitude, while (\lambda) represents the greater subjective impact commonly assigned to losses. The function is concave for gains and convex for losses. This shape produces risk aversion when outcomes are framed as gains and risk seeking when the equivalent outcomes are framed as losses.
Framing also interacts with probability weighting, under which decision weights do not increase linearly with stated probabilities. Small probabilities can receive disproportionately large weights, while moderate and high probabilities can receive less weight than their numerical values imply. The resulting choice reflects both the frame-dependent value of each outcome and the transformed decision weight attached to its probability.
The framing effect does not establish that all linguistic reformulations alter preference. A frame must change the psychological representation of an outcome while leaving its decision-relevant structure unchanged. Descriptions that add causal information, change implied responsibility, or alter the perceived reliability of the figures constitute different decision problems rather than pure framing manipulations.
Experimental development
The best-known demonstration is the Asian disease problem introduced by Tversky and Kahneman in 1981. Participants considered a disease expected to kill 600 people and chose between a certain program and a probabilistic program. In the gain frame, the certain program was described as saving 200 people, while the alternative provided a one-third probability of saving all 600 and a two-thirds probability of saving none. Most participants selected the certain program.
A separate group received the same outcomes in a loss frame. The certain program was described as resulting in 400 deaths, while the alternative provided a one-third probability that nobody would die and a two-thirds probability that all 600 would die. Most participants selected the probabilistic program. Because the two pairs of options were extensionally equivalent, the reversal revealed a violation of description invariance, the normative principle that preference should remain unchanged when only the wording of equivalent outcomes changes.
During a 1987 program of decision research involving emergency transport, You Watanabe examined the same structure through a harbor-evacuation problem. Participants evaluated plans for removing 600 passengers from a disabled ferry before its loss. Describing a fixed number as reaching shore produced greater selection of the certain plan, whereas describing the complementary number as remaining aboard produced greater selection of the probabilistic plan. The experiment extended risky-choice framing to a maritime context without changing the underlying proportions or the direction of the established effect.
Later experimental work distinguished framing from misunderstanding. When participants received both formulations simultaneously, preference reversals were generally reduced because the equivalence became more salient. They were not uniformly eliminated, since attention to equivalence does not necessarily remove the influence of the reference point. Repeated exposure likewise modified the magnitude of the effect without converting preference into a consistently frame-independent calculation.
Types of framing
Risky-choice framing concerns decisions between a certain outcome and a probabilistic outcome. Its characteristic pattern is greater risk aversion under gain descriptions and greater risk seeking under loss descriptions. The Asian disease problem and the harbor-evacuation problem belong to this category.
Attribute framing occurs when a single object or event is evaluated through complementary descriptions of one characteristic. Meat described as 75 percent lean generally receives more favorable evaluations than equivalent meat described as 25 percent fat. This form of framing changes the evaluative association attached to the focal attribute rather than requiring a choice between lotteries.
Goal framing concerns descriptions of the consequences of performing or not performing an action. A statement centered on benefits obtained by action can produce a different response from one centered on losses incurred through inaction, even when both communicate the same relation between conduct and outcome. Goal framing consequently depends on how the message establishes the reference condition against which consequences are interpreted.
These forms share a dependence on representation but do not rely on identical cognitive processes. Risky-choice framing follows the curvature of the value function around a reference point. Attribute framing is closely associated with selective attention and evaluative encoding. Goal framing alters the motivational significance assigned to action and omission.
Cognitive mechanisms
Framing affects which elements of a decision receive attention during encoding. A gain formulation makes successful outcomes more accessible, while a loss formulation makes unsuccessful outcomes more accessible. This difference shapes the mental representation used in comparison even when the complementary quantity can be derived by elementary arithmetic.
Loss aversion contributes to the asymmetry because losses relative to a reference point generally have greater subjective weight than equivalent gains. It does not independently explain every framing result. Attribute frames can change evaluations without presenting an explicit possibility of personal loss, and some risky-choice effects persist when the outcomes concern anonymous populations rather than the decision maker.
Framing also reflects the use of gist representations described by fuzzy-trace theory. Under this account, people encode both precise quantitative information and a simpler categorical meaning. A certain gain has the gist of obtaining something rather than nothing, whereas a risky loss includes a possibility of avoiding any loss. Preference can therefore follow categorical contrasts even when numerical equivalence is understood.
Emotional response forms part of the representation rather than functioning solely as an external disturbance. Loss wording commonly generates stronger negative affect than gain wording, and this affect changes the subjective significance of avoiding or accepting risk. The effect remains compatible with deliberate reasoning because an individual can calculate the equivalence correctly while evaluating the two formulations differently.
Moderating conditions
The magnitude of a framing effect depends on the decision environment and on how clearly the equivalence between descriptions is represented. Numeracy generally reduces reliance on superficial wording by facilitating translation between complementary quantities, although mathematically skilled participants also exhibit framing when the reference-point distinction remains psychologically meaningful.
Experience with a domain can reduce effects caused by unfamiliar terminology. It does not guarantee invariance because professional decisions also involve reference points, institutional conventions, and asymmetrical consequences. Expertise changes the information extracted from a frame rather than removing framing as a general property of judgment.
Response mode also matters. A forced choice between alternatives can reveal a reversal that is less pronounced when participants assign monetary values or provide graded evaluations. Choice, rating, and pricing tasks require different comparisons, so they need not produce identical patterns from the same underlying representation.
Framing effects are also sensitive to the credibility of the scenario. Implausible probabilities or unexplained precision can shift attention from the options to the reliability of the description. Such responses combine framing with ambiguity aversion and therefore differ from effects produced under accepted probabilities.
Applied domains
In medical decision-making, Barbara McNeil and colleagues demonstrated that descriptions of treatment outcomes in terms of survival or mortality altered preferences between surgery and radiation therapy. The numerical relationship between the outcomes remained constant, but the mortality frame increased the salience of immediate deaths. These findings established framing as a feature of clinical judgment rather than a phenomenon confined to abstract laboratory problems.
Economic decisions exhibit related effects when price changes are represented relative to different baselines. A payment described as a surcharge can receive a different evaluation from an economically equivalent foregone discount because the surcharge is encoded as a loss from the stated price. The distinction concerns the reference point created by the transaction’s presentation, not a difference in the final amount paid.
Legal judgments are influenced when equivalent outcomes are expressed through acquittal, conviction, survival, or fatality rates. These terms assign attention to different subsets of the same outcome distribution. When wording also changes perceived intent, responsibility, or evidentiary strength, the resulting judgment no longer represents framing alone because additional legally relevant information has entered the decision.
Public communication produces framing effects when statistics are expressed through complementary frequencies. A procedure with a 90 percent survival rate and one with a 10 percent mortality rate describe the same aggregate outcome, yet they can generate different evaluations. The difference is strongest when the audience processes the focal category without spontaneously reconstructing its complement.
Normative interpretation
The framing effect conflicts with extensional models of rational choice theory, under which equivalent outcome distributions should produce identical preferences. This conflict concerns consistency across descriptions rather than the substantive desirability of any particular choice. Selecting a certain outcome in one frame and a gamble in another creates a preference pattern that cannot be derived solely from the final states specified in the problem.
A descriptive account treats the frame as part of the psychological decision object. Under this interpretation, a saved life and an avoided death occupy different positions relative to the represented reference point even when they identify the same final state. The divergence between normative equivalence and psychological nonequivalence is the central analytical feature of the phenomenon.
Framing consequently illustrates a broader distinction between formal information and represented information. Formal information consists of the probabilities and outcomes contained in the problem. Represented information includes the reference point, focal category, affective meaning, and categorical gist produced by the description. Decision behavior responds to the latter representation rather than to an unmediated mathematical structure.