29.05.02 · psychology / cognition-intelligence

Decision-making and judgment: heuristics, biases (Kahneman-Tversky), dual-process theory

stub3 tiersLean: nonepending prereqs

Anchor (Master): Kahneman, D. — Thinking, Fast and Slow (2011)

Intuition Beginner

Daniel Kahneman won the Nobel Prize for showing that human decision-making is far from rational. Our brains use mental shortcuts called heuristics. They usually work, but they produce predictable errors in tricky situations. This unit is about those shortcuts, the errors they cause, and what they reveal about the mind.

The availability heuristic makes us overestimate dangers we hear about often — plane crashes — and underestimate common ones like car accidents. The representativeness heuristic makes us judge by stereotypes rather than statistics. Anchoring lets the first number mentioned distort our judgment: a $100 s hi r t ma r k e dd o w n t o$ 50 feels like a deal, even if $50 is still expensive.

Kahneman proposed two thinking systems. System 1 is fast, automatic, and intuitive. System 2 is slow, effortful, and deliberate. Most of the time we run on System 1, which works well enough but produces predictable errors when a situation is tricky. Knowing when to switch to System 2 is the skill this unit develops.

Visual Beginner

Figure: The heuristics-and-biases program at a glance. Three heuristics generate a family of predictable biases; prospect theory's S-shaped value function replaces the rational straight line with a curve that is steeper for losses; and dual-process theory frames the whole as a negotiation between a fast intuitive system and a slow deliberative one.

Worked example Beginner

The Asian disease problem

Tversky and Kahneman posed a hypothetical. A disease threatens 600 people. You must choose between two programs. Program A saves 200 people for certain. Program B gives a one-third chance of saving all 600 and a two-thirds chance of saving nobody. Asked this way, most people pick A — the certain rescue.

Now the same problem, reframed. Program C means 400 people die. Program D gives a one-third chance nobody dies and a two-thirds chance all 600 die. Asked this way, most people now pick D — the gamble.

The two framings are mathematically identical: saving 200 of 600 is the same as 400 dying. Yet the choice flips. This is the framing effect, the signature finding of prospect theory. People are risk-averse for gains and risk-seeking for losses, and "gain" versus "loss" is set entirely by the wording.

Check your understanding Beginner

Formal definition Intermediate

The vocabulary of judgment and decision-making is standardised across the anchor texts ^{[source pending]}. These terms identify empirically dissociable mechanisms, not mere labels.

Heuristics (Tversky & Kahneman, 1974) are cognitive shortcuts that substitute an easy question for a hard one. Instead of computing an actual probability, the mind asks how easily examples come to mind (availability); instead of consulting base rates, it asks how well an instance matches a prototype (representativeness); instead of estimating from scratch, it adjusts from a suggested starting value (anchoring and adjustment). Each substitution is adaptive in the sense that it trades accuracy for speed, and each generates a characteristic family of biases when the easy question and the hard question come apart.

Heuristic	Substituted question	Characteristic biases
Availability	"How easily do examples come to mind?"	Overweighting vivid/recallable events; underestimating common causes
Representativeness	"How much does this resemble a prototype?"	Base-rate neglect, conjunction fallacy, gambler's fallacy
Anchoring & adjustment	"Adjust from this starting value"	Insufficient adjustment; dragged estimates

Cognitive biases are systematic, predictable deviations from a normative standard, not random errors. The biases central to this unit include the framing effect (equivalent outcomes are evaluated differently depending on whether they are described as gains or losses — a surgery described as "90% survival" is chosen more often than the identical surgery described as "10% mortality"), hindsight bias (past events seem predictable in retrospect — "I knew it all along"), confirmation bias (selective search for and weighting of evidence that supports one's existing beliefs), the sunk cost fallacy (continuing a failing venture because of past, irrecoverable investment), the endowment effect (valuing owned goods more than identical goods not owned), status quo bias (preferring the current state to any change), the overconfidence effect (overestimating the accuracy of one's own knowledge and predictions), the gambler's fallacy (expecting independent events to "balance out"), and the conjunction fallacy (judging a conjunction $A \cap B$ more probable than its conjunct $A$ ).

Prospect theory (Kahneman & Tversky, 1979) is the descriptive model of choice under risk that displaced expected-utility theory as an account of how people actually decide. Its two central components are a value function $v$ defined over gains and losses relative to a reference point, and a probability weighting function $π$ that transforms objective probabilities into decision weights. The overall value of a prospect is

V = i \sum π (p_{i}) v (x_{i}),

where $x_{i}$ are outcomes measured as deviations from the reference point and $p_{i}$ their probabilities. The value function has three signature properties: it is defined on gains and losses rather than final wealth (reference dependence); it is concave for gains and convex for losses ( $v^{''} (x) < 0$ for $x > 0$ , $v^{''} (x) > 0$ for $x < 0$ ), producing risk aversion in the gain domain and risk seeking in the loss domain; and it is steeper for losses than for gains — loss aversion, with the loss-aversion coefficient $λ$ empirically estimated near $2.25$ , meaning losses hurt roughly twice as much as equivalent gains feel good. The weighting function $π$ overweights small probabilities (which is why the same person may both buy lottery tickets and buy insurance) and exhibits subcertainty, producing the certainty effect — the extra weight placed on outcomes that are guaranteed rather than merely probable.

Dual-process theory partitions cognition into two systems. System 1 is fast, automatic, parallel, effortless, largely unconscious, intuitive, and emotional. System 2 is slow, serial, effortful, deliberate, and tied to working memory. System 1 continuously generates impressions and intuitive judgments; System 2 monitors and endorses, revises, or overrides them — but only when sufficiently motivated and resourced. Because System 2 is metabolically costly to run, many System 1 outputs are accepted as-is, which is why biases are so robust and so resistant to mere awareness.

Key model Intermediate

Prospect theory as the unifying model

The 1979 paper "Prospect Theory: An Analysis of Decision under Risk" is the single most influential model in behavioral decision research, and it earns that status by subsuming a large catalogue of biases under three parametric assumptions about the value function and one about probability weighting ^{[source pending]}.

Reference dependence. Carriers of value are gains and losses relative to a reference point, not final states of wealth. A salary move from $$60, 000$ to $$55, 000$ is experienced as a loss; a move from $$50, 000$ to $$55, 000$ is experienced as a gain, even though the final state is identical. The reference point is usually the status quo but can be an expectation, an aspiration level, or a social comparison, and it is itself manipulable by framing — which is why the Asian disease problem flips choices.

Diminishing sensitivity. The value function is concave for gains and convex for losses. This produces risk aversion over gains (the agent prefers a certain $$500$ to a $50%$ chance of $$1, 000$ ) and risk seeking over losses (the agent prefers a $50%$ chance of losing $$1, 000$ to a certain loss of $$500$ ). The two preferences are mirror images of a single curvature change across the reference point — the reflection effect.

Loss aversion. The function is steeper for losses than for gains, $v (- x) = - λ v (x)$ with $λ \approx 2.25$ . Losing $$100$ hurts roughly twice as much as gaining $$100$ feels good. Loss aversion explains the endowment effect (ownership shifts the reference point so that selling feels like a loss), the status quo bias (any change away from the reference point is weighted by the prospect value function, which is convex-concave and loss-averse), and the asymmetry of negotiated concessions.

Probability weighting. The weighting function $π (p)$ overweights small probabilities and underweights moderate-to-large probabilities, with subcertainty ( $π (p) + π (1 - p) < 1$ ) that produces the certainty effect. The four-fold pattern of risk attitudes falls out directly: people are risk-averse for high-probability gains and low-probability losses (buying insurance against rare catastrophes) and risk-seeking for low-probability gains (lottery tickets) and high-probability losses (gambling to avoid a likely loss).

What the model explains. A single architecture — reference-dependent, loss-averse, S-shaped value combined with inverse-S probability weighting — accounts for the framing effect, the certainty effect, the reflection effect, the common-ratio and common-consequence violations of expected utility, and the four-fold pattern. Few models in psychology unify this many phenomena under this few parameters.

What the model does not explain. Original prospect theory violates stochastic dominance when more than two outcomes are involved, a defect repaired by cumulative prospect theory (Tversky & Kahneman, 1992), which applies the weighting function to cumulative probabilities — rank-dependent utility — with separate weighting curves for gains and losses. Prospect theory is also silent on the source of the reference point, which experimental work shows is context-dependent, and on choice under ambiguity (where probabilities are unknown rather than merely stated), which requires additional machinery developed in the master tier below.

Exercises Intermediate

Exercise 1 (medium, multiple choice).

A wine enthusiast bought a bottle for $20. I t i s n o w w or t h$ 200 on the resale market, but he refuses to sell it for less than $300— t h o ug hh e w o u l d n e v er p a y$ 300 to buy the same bottle. Which combination of prospect-theory concepts best explains this?

A. Diminishing sensitivity only B. Loss aversion and the endowment effect C. Probability weighting and the certainty effect D. Reference independence and risk seeking

Hint

Ownership shifts what counts as the reference point. Think about how willingness-to-accept and willingness-to-pay differ under loss aversion.

Answer

Option B. Ownership shifts the reference point so that selling feels like a loss, and loss aversion inflates the minimum acceptable price above the buying price. This gap between willingness-to-accept and willingness-to-pay is the endowment effect. Diminishing sensitivity alone cannot produce the buying/selling asymmetry, and the other options misidentify the mechanism.

Exercise 2 (medium, short answer).

You are given the choice between (A) a certain gain of $240 an d (B) a 25$ 1,000. Most people choose A. You are then given the choice between (C) a certain loss of $750 an d (D) a 75$ 1,000. Most people choose D. Using prospect theory, explain in 3-4 sentences why the risk attitude reverses between the two choices, and identify which property of the value function drives the reversal.

Hint

The reference point is the same in both choices. What changes is the sign of the outcomes, and therefore which side of the value function you are on.

Answer

The reversal is the reflection effect. In the gain domain, the value function is concave, so the agent prefers the certain gain — risk aversion. In the loss domain, the value function is convex, so the agent prefers the gamble — risk seeking. The single property driving both is the change in curvature of $v$ across the reference point: concave for gains, convex for losses. Expected utility theory, with its single global curvature, cannot reproduce both attitudes for the same agent without contradiction, which is the core empirical motivation for prospect theory.

Exercise 3 (medium, multiple choice).

The "Linda" problem (a philosophy major, outspoken, concerned with discrimination — is she more likely to be "a bank teller" or "a bank teller and active in the feminist movement"?) elicits the conjunction fallacy because:

A. People compute probabilities correctly but report them incorrectly. B. System 1 matches Linda to the feminist stereotype rather than applying the rule that a conjunction cannot exceed the probability of either conjunct. C. The description makes the conjunction objectively more probable. D. Loss aversion inflates the perceived frequency of feminists.

Answer

Option B. System 1 judges by representativeness — how well each option fits Linda's description — rather than by the probability calculus, which guarantees $P (A \cap B) \leq P (A)$ . Adding the feminist detail makes the conjunction fit the stereotype better, so it feels more probable, even though it is a strict subset. The fallacy persists even among statistically trained respondents, showing that representativeness can override explicit probability knowledge.

Exercise 4 (hard, application).

A government wants to increase the rate at which employees enroll in a retirement savings plan. Drawing on the framing effect, default options (choice architecture), and loss aversion, propose a concrete intervention and explain in 4-5 sentences why each design choice works.

Hint

Most people never change a pre-set option, and losses feel roughly twice as costly as equivalent gains. Combine both.

Answer

Make enrollment the default, with employees actively opting out rather than opting in, and describe contributions as "keeping your full employer match" (a loss frame for not contributing) rather than "gaining a match" (a gain frame). The default exploits status quo bias: most people never change a pre-set option, so enrollment rises without coercion. The loss frame exploits loss aversion: framing the foregone match as a loss makes non-participation feel roughly twice as costly as the equivalent gain frame, raising the perceived cost of inaction. Together these are a "nudge" in the Thaler-Sunstein sense — choice architecture that steers behavior without restricting options — and they work precisely because they are calibrated to the biases this unit catalogues.

Advanced results Master

Expected utility theory and its violations

The normative benchmark against which prospect theory is measured is expected utility (EU) theory, axiomatised by von Neumann and Morgenstern (1944). An agent satisfying four axioms — completeness, transitivity, continuity, and the independence axiom — acts as if maximising the expected value of a utility function over final wealth. The independence axiom is the load-bearing one: if you prefer gamble $A$ to gamble $B$ , then for any probability $p$ and any gamble $C$ , you must prefer the mixture $p A + (1 - p) C$ to $pB + (1 - p) C$ .

Allais paradox (1953). Maurice Allais constructed two choice pairs in which a majority of respondents violate the independence axiom. The common-consequence structure exposes certainty-seeking for gains: people prefer a certain large gain to a probabilistic one, but when both options are mixed with the same probabilistic outcome, the preference reverses. Allais's paradox was the first systematic demonstration that actual choices do not satisfy the EU axioms, and it directly motivated prospect theory's certainty effect and its non-linear probability weighting.

Ellsberg paradox (1961). Daniel Ellsberg showed that people prefer bets with known probabilities to bets with unknown (ambiguous) probabilities, even when the known-probability bet is objectively no better. This ambiguity aversion cannot be captured by expected utility, which assumes that subjective probabilities exist and are well-defined, and it motivates models of decision under ambiguity rather than risk — including Schmeidler's Choquet expected utility and Gilboa-Schmeidler max-min expected utility. The distinction between risk (probabilities known) and ambiguity (probabilities unknown) is foundational to modern decision theory.

Cumulative prospect theory

Original prospect theory (1979) applies a single probability weight per outcome and thereby violates stochastic dominance for prospects with three or more outcomes. Cumulative prospect theory (Tversky & Kahneman, 1992) repairs this by applying the weighting function to cumulative probabilities — rank-dependent utility — with separate weighting functions for gains and losses. The model retains the S-shaped value function and loss aversion while accommodating the full range of observed violations, and it supplies the empirically estimated parametric forms — $v (x) = x^{α}$ for gains, $v (x) = - λ (- x)^{β}$ for losses, with $α \approx β \approx 0.88$ and $λ \approx 2.25$ — that anchor most modern behavioral-economics applications.

Behavioral economics and choice architecture

The translation of prospect theory into policy is behavioral economics and its applied arm, choice architecture. Richard Thaler and Cass Sunstein's Nudge (2008) argues that the way options are presented — defaults, framing, ordering, salience — systematically shapes choice, and that policymakers can design choice environments that steer people toward better outcomes without restricting their freedom. Automatic enrollment in retirement plans, opt-out organ donation, and simplified menu ordering all exploit the status quo bias and the framing effect catalogued above. The empirical evidence is substantial: automatic enrollment raises retirement-savings participation dramatically, and defaults in end-of-life care and organ donation move population-level rates by tens of percentage points. The ethical boundary — when a nudge becomes manipulation — remains contested, and turns on transparency, the ease of opting out, and whether the nudge serves the chooser's interests or the architect's.

Neuroeconomics

Cognitive neuroscience has begun to map the value-computation machinery prospect theory describes. The ventromedial prefrontal cortex (vmPFC) encodes an integrated value signal that tracks subjective value under prospect-theoretic, not expected-utility, weighting — including reference dependence and loss aversion. The anterior insula activates in anticipation of losses and tracks loss-aversion-related arousal. The dopaminergic midbrain encodes a reward prediction error $δ_{t} = r_{t} + V (s_{t + 1}) - V (s_{t})$ , the discrepancy between expected and obtained reward, which is the learning signal of reinforcement-learning models and connects prospect theory to temporal-difference learning. The nontrivial neuroscientific finding is that the brain does not compute expected utility; it computes something much closer to prospect value, which is why the behavioral violations are so robust — they are wired in at the valuation stage.

Moral decision-making: Greene's dual-process theory

Joshua Greene extended dual-process theory to moral judgment. Using trolley-problem variants, he showed that personal moral dilemmas (pushing a large person off a bridge to stop a trolley) activate emotional regions (medial prefrontal cortex, amygdala) and tend to produce deontological judgments ("do not push"), while impersonal dilemmas (flipping a switch to divert the trolley) activate cognitive regions (dorsolateral prefrontal cortex) and tend to produce utilitarian judgments ("maximise lives saved"). On Greene's model, deontological judgments arise from emotional System 1 responses, while utilitarian judgments arise from cognitive System 2 override. The model is contested — some moral philosophers reject the equation of emotion with deontology — but it is the most developed neural-level dual-process account of moral cognition.

Intertemporal choice and hyperbolic discounting

Decisions across time exhibit hyperbolic discounting: future rewards are discounted at a rate inversely proportional to delay, $V = A / (1 + k d)$ , rather than the constant rate of exponential discounting $V = A e^{- r d}$ assumed by standard economics. Hyperbolic discounting produces time inconsistency — a preference reversal in which an agent prefers a smaller-sooner reward when choosing now but a larger-later reward when choosing in advance — which is the formal signature of present bias and self-control failure. David Laibson's quasi-hyperbolic ( $β$ - $δ$ ) model captures this with a present-bias parameter $β < 1$ applied to all future periods. Intertemporal choice connects to prospect theory through the role of the reference point (today's consumption) and loss aversion over delayed outcomes.

The hot-cold empathy gap

George Loewenstein's hot-cold empathy gap is a self-prediction failure: people in a "cold" (calm, unaroused) state systematically underestimate how they will behave in a "hot" (hungry, angry, sexually aroused, pain-afflicted) state. Smokers, dieters, and shoppers all report intentions in cold states that they violate in hot states. The empathy gap is a metacognitive bias — a bias about one's own future biases — and it connects dual-process theory to self-control: System 2 cannot effectively regulate System 1 if it cannot predict System 1's strength.

Ecological rationality: Gigerenzer's program

Gerd Gigerenzer has mounted the most sustained critique of the heuristics-and-biases program. On his ecological rationality view, heuristics are not error-producing shortcuts but adaptive tools matched to the structure of their environments. A heuristic is "fast and frugal" — it ignores most available information and uses a few simple cues — and it succeeds precisely because real environments are structured so that a few cues carry most of the predictive weight. The recognition heuristic (choose the option you recognise over the one you do not) beats regression models at predicting, for example, which of two cities is larger, when recognition correlates with the criterion. Gigerenzer's adaptive toolbox catalogs dozens of such heuristics, each matched to an environmental structure. The disagreement with Kahneman is less about the data — both camps agree heuristics exist and produce characteristic patterns — than about interpretation: Kahneman frames heuristics as departures from a normative standard; Gigerenzer frames them as matching to environmental structure. Both are defensible, and the most productive work treats the two as complementary rather than opposed.

Recognition-primed decisions: expert intuition

Gary Klein's recognition-primed decision (RPD) model, developed from studies of firefighters, military commanders, and nurses, describes how experts make decisions under time pressure. Rather than comparing multiple options, the expert recognises the situation as typical, retrieves a default action, and mentally simulates whether it will work — revising only if the simulation fails. RPD is not the elimination of System 2; it is the compression of System 2 into a rapid simulation grounded in extensive System 1 pattern matching. The model dovetails with the ecological-rationality view: expert intuition works when the environment provides reliable cues and the expert has learned them through feedback. It fails when cues are unreliable or feedback is absent — which is why expert intuition is trustworthy in chess and firefighting but notoriously poor in stock-picking and psychiatric diagnosis.

The nontrivial upshot

The decision-making literature converges on a structural claim: human judgment is neither uniformly rational nor uniformly irrational. It is the output of bounded, adaptive systems whose accuracy depends on the match between the heuristic deployed and the structure of the environment. Prospect theory, dual-process theory, ecological rationality, and neuroeconomics are four angles on the same underlying architecture — an architecture that is fast, frugal, and systematically biased in predictable directions, and that no current formal model captures in full.

Connections Master

Cognition and intelligence 29.05.01 is the direct prerequisite and the parent unit. The heuristics, biases, and dual-process material introduced there is specialised here into full decision-theoretic detail, and prospect theory is developed as this unit's central model.
Statistical reasoning 29.01.03 pending supplies the normative benchmark. The base-rate fallacy, the conjunction fallacy, and the gambler's fallacy are all violations of the probability calculus formalised there, and the Bayesian framework is the standard against which human judgment is measured.
Neuroscience [29.02.NN] (pending) provides the substrate. The vmPFC value signal, the insula's loss-anticipation response, and the dopaminergic prediction error are the neural realisations of the valuation machinery prospect theory describes.
Learning and memory 29.04.01 connects through the role of availability — retrieval fluency as a cue — and through reinforcement learning, whose temporal-difference prediction error is the learning-theoretic counterpart of the neuroeconomic value signal.
Social psychology [29.07.NN] (pending) extends these mechanisms into the interpersonal domain: the framing of persuasive messages, the anchoring of social norms, and the role of group conformity in amplifying individual biases.
Psychological disorders [29.09.NN] (pending) inherits the valuation machinery: depression and anxiety as distortions of probability weighting and loss aversion, and addiction as a hyperbolic-discounting failure of self-control.
Motivation and emotion [29.11.NN] (pending) supplies the affective dimension. The hot-cold empathy gap and the role of emotion in moral judgment both depend on the interplay between affective and deliberative systems catalogued here.
Language and thought 29.05.03 pending (proposed successor) connects through framing: the wording of a prospect determines its reference point, and the heuristics-and-biases program provides the empirical foundation for studying how linguistic framing shapes judgment and communication.

Historical & philosophical context Master

The heuristics-and-biases program was created by two Israeli psychologists, Amos Tversky and Daniel Kahneman, whose collaboration began in the late 1960s and produced the most influential body of work in postwar psychology. Their 1974 Science paper, "Judgment under Uncertainty: Heuristics and Biases," catalogued the three canonical heuristics — availability, representativeness, anchoring — and the biases each generates. Their 1979 Econometrica paper, "Prospect Theory: An Analysis of Decision under Risk," supplied the descriptive model that displaced expected-utility theory as the standard account of choice under risk and inaugurated behavioral economics. Kahneman was awarded the Nobel Prize in Economics in 2002; Tversky had died in 1996 and could not share it. The Econometrica paper is among the most-cited works in the social sciences.

The deeper intellectual context is the critique of the rational-actor model that dominated postwar economics. Expected-utility theory, game theory, and rational-choice theory all assume that agents are consistent, well-informed utility maximisers. The empirical case for this assumption was always weaker than its formal elegance suggested: Herbert Simon's bounded rationality (1955) argued that real agents lack the cognitive and informational resources to optimise, and satisfice instead. Tversky and Kahneman made the critique quantitative: they showed not only that people deviate from the rational model but that the deviations are systematic, directional, and predictable — which is exactly what makes them both scientifically interesting and practically exploitable. A random error is noise; a systematic bias is a discoverable law.

The philosophical weight of the program is epistemological. If human reasoning is systematically biased, then the reliability of everyday judgment — including the judgments on which science, law, medicine, and policy depend — is in question. This does not lead to skepticism, because the biases are themselves knowable and correctable. It leads to a calibrated view of human rationality: we are reasoning animals whose reasoning is good enough most of the time and predictably bad in identifiable circumstances, and the identification of those circumstances is itself a cognitive achievement.

The Gigerenzer-Kahneman debate is the program's main internal controversy. Kahneman treats heuristics as departures from a normative (Bayesian or expected-utility) standard; Gigerenzer treats them as adaptive matches to environmental structure. The disagreement is partly empirical (which framing better predicts behavior in which task) and partly philosophical (what counts as the relevant normative standard). The mature position is that both lenses illuminate real phenomena, and that a complete theory of judgment will need both: an account of when heuristics match their environments and an account of when they misfire.

The translation into policy — behavioral economics, nudge, choice architecture — is the program's most visible legacy. Thaler's 2017 Nobel Prize, following Kahneman's, recognised that the descriptive theory of biased choice can be turned into prescriptive interventions that improve outcomes at population scale. The ethical question — who decides which direction is "better," and how transparent the architecture must be — is now a live topic in its own right, and it cannot be settled by the psychology alone. The science supplies the levers; the ethics must decide how to pull them.

Bibliography Master

Kahneman, D. & Tversky, A. — "Prospect theory: an analysis of decision under risk", Econometrica 47, 263–291 (1979).
Tversky, A. & Kahneman, D. — "Judgment under uncertainty: heuristics and biases", Science 185, 1124–1131 (1974).
Tversky, A. & Kahneman, D. — "Advances in prospect theory: cumulative representation of uncertainty", Journal of Risk and Uncertainty 5, 297–323 (1992).
Kahneman, D. — Thinking, Fast and Slow (Farrar, Straus and Giroux, 2011).
von Neumann, J. & Morgenstern, O. — Theory of Games and Economic Behavior (Princeton University Press, 1944).
Allais, M. — "Le comportement de l'homme rationnel devant le risque: critique des postulats et axiomes de l'école américaine", Econometrica 21, 503–546 (1953).
Ellsberg, D. — "Risk, ambiguity, and the Savage axioms", Quarterly Journal of Economics 75, 643–669 (1961).
Simon, H. A. — "A behavioral model of rational choice", Quarterly Journal of Economics 69, 99–118 (1955).
Thaler, R. H. & Sunstein, C. R. — Nudge: Improving Decisions About Health, Wealth, and Happiness (Yale University Press, 2008).
Gigerenzer, G., Todd, P. M. & the ABC Research Group — Simple Heuristics That Make Us Smart (Oxford University Press, 1999).
Klein, G. — Sources of Power: How People Make Decisions (MIT Press, 1998).
Greene, J. D., Sommerville, R. B., Nystrom, L. E., Darley, J. M. & Cohen, J. D. — "An fMRI investigation of emotional engagement in moral judgment", Science 293, 2105–2108 (2001).
Laibson, D. — "Golden eggs and hyperbolic discounting", Quarterly Journal of Economics 112, 443–477 (1997).
Loewenstein, G. — "Hot-cold empathy gaps and medical decision making", Health Psychology 24 (4 Suppl), S49–S56 (2005).
Stanovich, K. E. — Rationality and the Reflective Mind (Oxford University Press, 2011).
Camerer, C., Loewenstein, G. & Prelec, D. — "Neuroeconomics: how neuroscience can inform economics", Journal of Economic Literature 43, 9–64 (2005).
Tversky, A. & Kahneman, D. — "Rational choice and the framing of decisions", Journal of Business 59, S251–S278 (1986).
Myers, D. G. & DeWall, C. N. — Psychology, 13th ed. (Worth, 2021), Ch. 9.
Gleitman, H., Gross, J. & Reisberg, D. — Psychology, 8th ed. (Norton, 2011), Ch. 8.