Daniel Kahneman

1.1 Axiomatic Foundations of Homo Economicus

Neoclassical microeconomics was established on the axiomatic foundation of Expected Utility Theory (EUT), synthesized by John von Neumann and Oskar Morgenstern in Theory of Games and Economic Behavior (1944), and generalized to subjective probability by Leonard J. Savage in The Foundations of Statistics (1954).

Let $\mathcal{X} = \{x_1, x_2, \dots, x_n\}$ denote a set of deterministic monetary consequences, and let $\mathcal{L}$ denote the set of simple lotteries over $\mathcal{X}$: $$L = (p_1, x_1; p_2, x_2; \dots; p_n, x_n), \quad p_i \ge 0, \quad \sum_{i=1}^n p_i = 1$$

Von Neumann and Morgenstern proved that if an agent's preference relation $\succeq$ over $\mathcal{L}$ satisfies four axioms, there exists a cardinal utility function $u: \mathcal{X} \to \mathbb{R}$, unique up to positive affine transformation, representing preferences via expected utility:

               VON NEUMANN-MORGENSTERN AXIOMATIC ARCHITECTURE
                                     │
         ┌───────────────────────────┼───────────────────────────┐
         ▼                           ▼                           ▼
  [ Completeness & ]          [ Continuity ]             [ Independence ]
  [  Transitivity  ]                 │                           │
  Total ordering over          Archimedean axiom;         Substitution axiom:
  all lotteries; no           no infinite joy or         $L \succeq M \iff$
  intransitive loops.          infinite agony.            $\alpha L + (1-\alpha)N \succeq \alpha M + (1-\alpha)N$
         │                           │                           │
         └───────────────────────────┼───────────────────────────┘
                                     ▼
                      EXPECTED UTILITY THEOREM
                      $U(L) = \sum_{i=1}^n p_i u(x_i)$

1. Completeness: For any $L, M \in \mathcal{L}$, either $L \succeq M$, $M \succeq L$, or both ($L \sim M$).
2. Transitivity: If $L \succeq M$ and $M \succeq N$, then $L \succeq N$.
3. Continuity: If $L \succeq M \succeq N$, there exists a unique $\alpha \in [0, 1]$ such that $\alpha L + (1-\alpha) N \sim M$.
4. Independence (Substitution Axiom): For all $L, M, N \in \mathcal{L}$ and $\alpha \in (0, 1]$: $$L \succeq M \iff \alpha L + (1-\alpha) N \succeq \alpha M + (1-\alpha) N$$

From these axioms follows the representation theorem: $$U(L) = \mathbb{E}_L[u(x)] = \sum_{i=1}^n p_i u(x_i)$$

1.2 Savage's Subjective Expected Utility (SEU) and Postulate P2

Savage (1954) extended EUT to uncertain states $s \in S$. The core postulate is Postulate P2 (The Sure-Thing Principle): If an agent prefers act $f$ to act $g$ when event $E$ obtains, and also prefers $f$ to $g$ when event $E$ does not obtain ($\neg E$), then the agent must prefer $f$ to $g$ unconditionally: $$f_E \succeq g_E \quad \text{and} \quad f_{\neg E} \succeq g_{\neg E} \implies f \succeq g$$

1.3 Neoclassical Behavioral Assumptions

Neoclassical equilibrium models (Arrow & Debreu, 1954; Fama, 1970) rested on three axioms of rational agency: 1. Asset Integration: Payoffs $x$ are evaluated strictly in terms of final total wealth states $W$: an agent accepts a gamble if and only if $\mathbb{E}[u(W + x)] > u(W)$. 2. Bayesian Updating: New data $D$ updates prior beliefs via Bayes' theorem: $$P(H \mid D) = \frac{P(D \mid H) P(H)}{P(D)}$$ 3. Risk Attitude as Wealth Curvature: Risk preference is governed exclusively by the second derivative $u''(W)$: concave utility ($u''(W) < 0$) indicates risk aversion ($R_A(W) = -u''(W)/u'(W)$), while convex utility indicates risk seeking.

Kahneman and Tversky established that human judgment under uncertainty replaces formal probability calculations with intuitive heuristics operating via attribute substitution: swapping a difficult statistical query for an accessible perceptual or associative query.

                      CANONICAL HEURISTICS & ATTRIBUTE SUBSTITUTION
                                            │
        ┌───────────────────────────────────┼───────────────────────────────────┐
        ▼                                   ▼                                   ▼
 [ Representativeness ]              [ Availability ]               [ Anchoring & Adjustment ]
 Target: $P(A \in B)$                Target: Frequency / Base Rate   Target: Numerical Quantity
 Substituted: $S(A, B)$              Substituted: Retrieval Fluency  Substituted: Anchor Accessibility
 Error: Base-Rate Neglect,           Error: Salience Bias,           Error: Insufficient Adjustment,
 Conjunction Fallacy, Small Samples  Media Distortion, Vividness     Assimilation to Random Input

2.1 The Representativeness Heuristic

2.1.1 Mechanism and Definition

Representativeness evaluates the probability that object $x$ belongs to class $C$, or that process $P$ generates event $E$, by the degree of similarity $S(x, C)$ between the instance and the prototype.

2.1.2 Base-Rate Neglect (Kahneman & Tversky, 1973)

Subjects evaluated descriptions of individuals drawn from a pool of 100 professionals (engineers and lawyers): * Base Rates: Manipulated across groups (70 engineers / 30 lawyers vs. 30 engineers / 70 lawyers). * Finding: When given descriptive sketches (e.g., Jack, who enjoys carpentry and mathematical puzzles), subjects produced identical probability estimates ($P \approx 0.90$) across both base-rate conditions: $$P(\text{Engineer} \mid \text{Jack}) \approx f(S(\text{Jack}, \text{Engineer})) \quad \text{independent of } P(\text{Engineer})$$ * Uninformative Condition: Given a blank sketch ("Dick shows high ability..."), subjects estimated $P = 0.50$, ignoring prior base rates ($0.70$ or $0.30$) required by Bayesian conditionalization.

2.1.3 The Law of Small Numbers (Tversky & Kahneman, 1971)

People expect small samples to mirror population parameters: 1. Gambler’s Fallacy: Treating independent Bernoulli trials as self-correcting systems (e.g., expecting tails after consecutive heads). 2. Sample Size Insensitivity: Failing to account for sampling variance ($\sigma_{\bar{x}} = \sigma / \sqrt{N}$), judging small and large hospitals equally likely to experience extreme sex ratio skews.

2.1.4 The Conjunction Fallacy and the Linda Problem (1983)

Tversky and Kahneman presented Linda: 31, single, outspoken, philosophy major, deeply concerned with discrimination and social justice: * $T$: Linda is a bank teller. * $T \land F$: Linda is a bank teller and active in the feminist movement.

                           THE CONJUNCTION VIOLATION

              ┌─────────────────────────────────────────────────┐
              │  Bank Tellers (T)                               │
              │                                                 │
              │         ┌─────────────────────────────┐         │
              │         │  Feminist Bank Tellers      │         │
              │         │  (T ∧ F)                    │         │
              │         │                             │         │
              │         └─────────────────────────────┘         │
              └─────────────────────────────────────────────────┘

              Extensional Probability:  P(T ∧ F) ≤ P(T)
              Heuristic Judgment:       P(T ∧ F) > P(T)   [85%+ of subjects]

By set inclusion, $P(T \land F) \le P(T)$. Yet over 85% of subjects—including statisticians—ranked $P(T \land F) > P(T)$ because the conjunction matched Linda’s representative prototype.

2.2 The Availability Heuristic

2.2.1 Mechanism and Fluency

Probability and frequency are evaluated by the cognitive ease with which instances are retrieved from memory.

2.2.2 Experimental Paradigms

1. Lexical Retrieval (1973): Subjects judged whether English letters ($K, L, N, R, V$) appear more often in the first or third position. Two-thirds estimated position 1, despite position 3 being twice as frequent. Memory indexes words by phonetic onset, creating retrieval fluency for initial letters.
2. Lethal Risk Judgments (Lichtenstein et al., 1978): Causes of death with vivid, media-amplified coverage (homicide, plane crashes, tornadoes) were overestimated up to 1,000-fold. Quiet, high-frequency killers (stroke, diabetes, stomach cancer) were underestimated up to 100-fold. Fluency of episodic recall directly substituted for demographic incidence.

2.3 Anchoring and Adjustment

2.3.1 Experimental Paradigm (1974)

Subjects witnessed a wheel of fortune rigged to land on 10 or 65, then estimated the percentage of African nations in the UN: * Median estimate for anchor 10: 25% * Median estimate for anchor 65: 45% ($p < 0.001$) The uninformative numeric prime pulled estimates by 20 percentage points.

2.3.2 Dual Underlying Mechanisms

1. Insufficient Adjustment (System 2): Deliberate adjustment begins at the anchor and moves toward truth, terminating prematurely at the edge of a plausibility boundary to conserve cognitive effort (Epley & Gilovich, 2001).
2. Selective Accessibility (System 1): The anchor primes anchor-consistent semantic features in associative memory, biasing subsequent hypothesis testing (Mussweiler & Strack, 1999).

In 1979, Kahneman and Tversky published "Prospect Theory: An Analysis of Decision under Risk" in Econometrica, formalized in 1992 as Cumulative Prospect Theory.

                      PROSPECT THEORY: THE TWO-PHASE PIPELINE

  [ Choice Set / Lotteries ]
              │
              ▼
    ┌───────────────────┐
    │   EDITING PHASE   │  --> Coding (Reference Point Identification)
    │                   │  --> Combination / Segregation / Cancellation
    └─────────┬─────────┘
              │
              ▼
    ┌───────────────────┐
    │ EVALUATION PHASE  │  --> Value Function: v(x) [S-Shaped, Loss Averse]
    │                   │  --> Decision Weights: w(p) [Inverted S-Curve]
    └─────────┬─────────┘
              │
              ▼
    [ Total Prospect Value: V(L) ]

3.1 Axiomatic Violations of Expected Utility

3.1.1 The Allais Paradox (Certainty Effect)

• Choice 1: \$1M sure ($A$) vs. (89% \$1M, 10% \$5M, 1% \$0) ($B$). Majority: $A \succ B$.
• Choice 2: (11% \$1M, 89% \$0) ($C$) vs. (10% \$5M, 90% \$0) ($D$). Majority: $D \succ C$. Under EUT, $A \succ B \implies 0.11 u(\$1\text{M}) > 0.10 u(\$5\text{M}) + 0.01 u(\$0)$, while $D \succ C \implies 0.10 u(\$5\text{M}) + 0.01 u(\$0) > 0.11 u(\$1\text{M})$, directly violating the Independence Axiom.

3.1.2 Rabin's Calibration Theorem (2000)

Rabin proved that if an expected utility maximizer rejects a 50/50 lose \$100 / win \$110 bet across baseline wealth levels, the required curvature forces them to reject a 50/50 lose \$1,000 / win \$1,000,000 gamble. Risk aversion over modest stakes cannot be explained by diminishing marginal utility of wealth; it requires reference-dependent loss aversion.

3.2 The Asymmetric S-Shaped Value Function $v(x)$

The value function is defined over changes relative to a reference point $x = W - r$:

$$v(x) = \begin{cases} x^\alpha & \text{for } x \ge 0 \\ -\lambda (-x)^\beta & \text{for } x < 0 \end{cases}$$

                                  Value v(x)
                                      ▲
                                      │              Concave in Gains
                                      │             (Risk Aversion)
                                      │              . - ~ * '
                                      │       . - ~ '
                                      │ . - ~ '
                 Losses               │                     Gains
        ──────────────────────────────┼──────────────────────────────► x
               - ~ '                  │ (Reference Point: x = 0)
         . - ~                        │
       /                              │
      /                               │
     /                                │
    │ Convex in Losses                │
    │ (Risk Seeking)                  │
    │ Loss Aversion:                  │
    │ Slope at 0⁻ is ~2.25x Slope at 0⁺
    ▼

3.2.1 Core Mathematical Properties

1. Reference Dependence: $v(0) = 0$. Value reflects deviations from baseline.
2. Diminishing Sensitivity: * Gains ($x > 0$): $v''(x) \le 0$ (concave $\to$ risk aversion). * Losses ($x < 0$): $v''(x) \ge 0$ (convex $\to$ risk seeking).
3. Loss Aversion: $$-v(-x) > v(x) \quad \text{for all } x > 0$$ Empirically (Tversky & Kahneman, 1992): $\alpha = \beta \approx 0.88$, and $\lambda \approx 2.25$. The function has a non-differentiable kink at the reference point: $$\lim_{x \to 0^+} \frac{v'(-x)}{v'(x)} = \lambda \approx 2.25$$

3.3 The Probability Weighting Function $w(p)$

Decision makers transform probabilities non-linearly via decision weights:

                           DECISION WEIGHTING FUNCTION w(p)

               Decision Weight w(p)
               1.0 ▲                                    . - ~ * Certainty
                   │                             . - ~ '      (p = 1)
                   │                      . - ~ '
                   │               . - ~ '  Underweighting of
                   │        . - ~ '         Moderate & High Probabilities
                   │ . - ~ '
   Overweighting   │/ . - - - - - - - - - -  Identity Line: w(p) = p
   of Tail Events  /
             0.0   └────────────────────────────────────────► Objective p
                  0.0                                      1.0

1. Overweighting Small Probabilities: $w(p) > p$ for $p < 0.1$, explaining simultaneous lottery participation and catastrophe insurance purchases.
2. Underweighting Moderate/High Probabilities: $w(p) < p$ for $p > 0.3$.
3. Parametric Form (Tversky & Kahneman, 1992): $$w^+(p) = \frac{p^\gamma}{\left(p^\gamma + (1-p)^\gamma\right)^{1/\gamma}}, \quad w^-(p) = \frac{p^\delta}{\left(p^\delta + (1-p)^\delta\right)^{1/\delta}}$$ Estimated medians: $\gamma \approx 0.61$ (gains), $\delta \approx 0.69$ (losses).
4. Prelec's Axiomatic Weighting (1998): $$w(p) = \exp\left(-\beta (-\ln p)^\alpha\right), \quad 0 < \alpha < 1, \quad \beta > 0$$

3.4 The Fourfold Pattern of Risk Attitudes

3.5 Cumulative Prospect Theory (1992)

Original 1979 Prospect Theory permitted violations of First-Order Stochastic Dominance (FOSD). Cumulative Prospect Theory resolved this by applying decision weights to the cumulative probability distribution (Quiggin, 1982).

For ordered outcomes $x_{-m} < \dots < x_0 = 0 < \dots < x_n$: $$V(L) = \sum_{i=-m}^n \pi_i v(x_i)$$ Where: $$\pi_i^+ = w^+\left(\sum_{j=i}^n p_j\right) - w^+\left(\sum_{j=i+1}^n p_j\right), \quad \pi_i^- = w^-\left(\sum_{j=-m}^i p_j\right) - w^-\left(\sum_{j=-m}^{i-1} p_j\right)$$ Because weights represent marginal cumulative differences, stochastic dominance is strictly preserved.

3.6 Framing and the Endowment Effect

1. Asian Disease Framing (1981): Framing an identical intervention as "200 saved" vs. "400 die" shifted choices from 72% risk-averse to 78% risk-seeking.
2. Endowment Effect (Kahneman, Knetsch, & Thaler, 1990): Cornell students endowed with mugs demanded a median selling price (WTA) of \$7.12, while buyers offered a median price (WTP) of \$2.87 ($\text{WTA}/\text{WTP} \approx 2.48 \approx \lambda$). Property ownership shifts the reference point, rendering surrender a loss.

In Thinking, Fast and Slow (2011), Kahneman formalized cognitive functioning into two interacting modes (Stanovich & West, 2000).

                        DUAL-PROCESS DICHOTOMY MATRIX

     SYSTEM 1 (Fast, Intuitive)                SYSTEM 2 (Slow, Deliberative)
  ┌──────────────────────────────────────┐  ┌──────────────────────────────────────┐
  │ • Associative & Narrative Coherence │  │ • Rule-Governed & Algorithmic        │
  │ • Fast, Parallel, Automatic          │  │ • Slow, Serial, Effortful            │
  │ • Involuntary (Cannot be turned off) │  │ • Controlled & Self-Monitored        │
  │ • Neglects Ambiguity & Evidence Void │  │ • Evaluates Conflicting Hypotheses   │
  │ • WYSIATI Engine                     │  │ • Working-Memory Constrained         │
  │ • Metabolic Cost: Low                │  │ • Metabolic Cost: High (Pupil Dilates│
  └──────────────────┬───────────────────┘  └──────────────────┬───────────────────┘
                     │                                         │
                     └───────────────────┬─────────────────────┘
                                         ▼
                 System 2 Defaults to "Lazy Endorsement";
                 Functions as Retrospective Defense Attorney.

4.1 Comparative Characteristics

4.2 Attribute Substitution & WYSIATI

System 1 constructs causal narratives from available cues. It suppresses alternative interpretations and ignores absent information (WYSIATI). Subjective confidence reflects narrative coherence, not statistical validity.

4.3 Cognitive Reflection Test (Frederick, 2005)

"A bat and ball cost \$1.10. The bat costs \$1.00 more than the ball. How much does the ball cost?" * System 1: \$0.10 (intuitive, fast, incorrect). * System 2: \$0.05 ($(x + 1.00) + x = 1.10 \implies x = 0.05$). * Failure to override System 1 reflects cognitive laziness: System 2 endorses intuitive proposals without computation.

4.4 Physiological Substrates & Pupillometry

Kahneman and Beatty (Science, 1966) demonstrated that pupillary diameter tracks cognitive workload: * Baseline: $\sim 3.0\text{ mm}$. * Memory load (Add-3 transformation under metronome cadence): dilation up to $+0.8\text{ mm}$ ($+25\%$). * Constriction occurs within 200 ms upon task termination or cognitive abandonment, establishing pupil dilation as an involuntary real-time index of System 2 mental strain.

Kahneman decoupled instantaneous experienced utility from retrospective remembered utility ("Back to Bentham?", 1997).

               THE SPLIT ARCHITECTURE OF SUBJECTIVE TIME
                                   │
         ┌─────────────────────────┴─────────────────────────┐
         ▼                                                   ▼
 [ EXPERIENCING SELF ]                               [ REMEMBERING SELF ]
 Temporal Reality: Continuous Time                   Temporal Reality: Atemporal Snapshot
 Calculus: Integral of Affect                        Calculus: Heuristic Average
 $U_{\text{exp}} = \int_0^T u(t) \, dt$              $U_{\text{rem}} \approx \frac{u_{\text{peak}} + u_{\text{end}}}{2}$
 Duration: Linear Weighting                          Duration: TOTAL NEGLECT

5.1 The Mathematical Model

Let $u(t)$ be affective valence at time $t \in [0, T]$: * Experiencing Self: $$U_{\text{experienced}} = \int_0^T u(t) \, dt$$ * Remembering Self (Peak-End Rule & Duration Neglect): $$U_{\text{remembered}} \approx \frac{\max_{t} |u(t)| + u(T)}{2}, \quad \frac{\partial U_{\text{remembered}}}{\partial T} \approx 0$$

5.2 Clinical & Experimental Evidence

1. Colonoscopy Trials (Redelmeier & Kahneman, 1996; 2003): Retrospective pain ratings correlated with peak pain ($r \approx 0.67$) and end pain ($r \approx 0.58$), but showed zero correlation with duration ($r \approx 0.03$). Prolonging the procedure by leaving the endoscope stationary for 3 minutes reduced retrospective pain and increased return rates by 43%.
2. Cold Pressor Experiment (1993): Subjects endured 60s at 14°C vs. 60s at 14°C plus 30s warming to 15°C. Sixty-five percent chose to repeat the longer trial, voluntarily accepting 30 additional seconds of pain to obtain a milder retrospective memory.
3. Affective Forecasting: The focusing illusion demonstrates that "nothing in life is as important as you think it is, while you are thinking about it."

This register audits the empirical status of Kahneman and Tversky’s core findings against large-scale preregistered replications and Kahneman’s 2012 open letter.

6.2 The Replication Crisis and Scientific Integrity

• The 2012 Open Letter: Kahneman warned social psychologists: "I see a train wreck looming... The special need for replication in this field is undeniable."
• The 2017 Public Retraction: Kahneman conceded error regarding Chapter 4 of Thinking, Fast and Slow: "I accepted the results of published studies without adequate scrutiny... I was too credulous, and the chapter is defective."
• Adversarial Collaboration: Kahneman pioneered adversarial collaboration with Gary Klein (2009, expertise vs. bias) and Matthew Killingsworth (2023, income and well-being), establishing a standard for resolving scientific conflict.

7.1 AI Alignment, RLHF, and Reward Hacking

               AI ALIGNMENT THROUGH THE KAHNEMAN LENS
                                  │
         ┌────────────────────────┴────────────────────────┐
         ▼                                                 ▼
 [ RLHF & The Remembering Self ]                   [ Attribute Substitution in Evaluators ]
 Human annotators score model                    Reward models substitute hard semantic
 transcripts via Peak-End heuristics:             truth with accessible surface features:
 • Eloquent apologies                            • Output length and verbosity
 • Flawless closing paragraphs                   • Sycophancy and agreement with user
 • Duration/Step neglect                         • Formatting polish and markdown tables

1. Remembering Self in RLHF: Human annotators rating AI generation trajectories display the peak-end rule and duration neglect. Models learn to generate articulate apologies and persuasive closings rather than correct reasoning, optimizing for the remembering self's retrospective score.
2. Attribute Substitution in Reward Models: Reward models substitute factual accuracy (computationally difficult) with surface fluency, length, and politeness (accessible). This creates reward hacking, where AI systems produce convincing hallucinations.
3. Loss-Averse Autonomous Control: Embedding asymmetric loss parameters ($\lambda \gg 2$) into autonomous vehicle trajectory planners prevents catastrophic low-probability edge cases.

7.2 LLM Temperature, Reasoning Chains, and Dual-Process Emulation

1. System 1 as Autoregressive Decoding: A transformer generating tokens in a single forward pass operates as System 1—fast, associative pattern completion driven by softmax probabilities: $$P(w_t \mid w_{
2. Chain-of-Thought (CoT) as System 2 Serialization: CoT prompting (Wei et al., 2022) and inference-time search architectures (Tree of Thoughts, o1/o3 reasoning models) emulate System 2: serializing computation, utilizing the context window as a working-memory buffer, and performing self-correction.

7.3 Risk Modeling in Quantitative Financial Engineering

1. Failure of Gaussian VaR: Value-at-Risk models assuming normal distributions and symmetric utility fail during market stress. Market participants governed by loss aversion ($\lambda \approx 2.25$) and tail overweighting ($w(p) \gg p$) engage in non-linear panic selling, producing volatility clustering and fat tails.
2. Myopic Loss Aversion and Equity Premium (Benartzi & Thaler, 1995): Investors evaluate portfolios frequently (myopia) and suffer loss aversion. The historical 6% equity premium compensates investors not for return variance, but for the psychological penalty of enduring frequent quarterly losses.

7.4 Choice Architecture, Interface Engineering, and Friction

1. Default Effects and System 2 Inertia: In organ donation systems, opt-in (Germany: 12%) versus opt-out (Austria: 99.98%) defaults determine participation (Johnson & Goldstein, 2003). System 2 defaults to the status quo to conserve effort.
2. Loss Aversion Exploitation: E-commerce interfaces deploy countdown timers and cart expiration warnings to trigger loss aversion.
3. Defensive UX: In safety-critical software systems, requiring operators to manually type resource names prior to irreversible actions (e.g., deleting a database) forces System 1 interruption and compels deliberate System 2 engagement.

8.1 Foundational Collaborative Papers (Kahneman & Tversky)

• Tversky, A., & Kahneman, D. (1971). "Belief in the law of small numbers." Psychological Bulletin, 76(2), 105–110.
• Kahneman, D., & Tversky, A. (1972). "Subjective probability: A judgment of representativeness." Cognitive Psychology, 3(3), 430–454.
• Kahneman, D., & Tversky, A. (1973). "On the psychology of prediction." Psychological Review, 80(4), 237–251.
• Tversky, A., & Kahneman, D. (1973). "Availability: A heuristic for judging frequency and probability." Cognitive Psychology, 5(2), 207–232.
• Tversky, A., & Kahneman, D. (1974). "Judgment under uncertainty: Heuristics and biases." Science, 185(4157), 1124–1131.
• Kahneman, D., & Tversky, A. (1979). "Prospect theory: An analysis of decision under risk." Econometrica, 47(2), 263–291.
• Tversky, A., & Kahneman, D. (1981). "The framing of decisions and the psychology of choice." Science, 211(4481), 453–458.
• Tversky, A., & Kahneman, D. (1983). "Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment." Psychological Review, 90(4), 293–315.
• Tversky, A., & Kahneman, D. (1992). "Advances in prospect theory: Cumulative representation of uncertainty." Journal of Risk and Uncertainty, 5(4), 297–323.

8.2 Books, Monographs & Experimental Physiology

• Kahneman, D., & Beatty, J. (1966). "Pupil diameter and load on memory." Science, 154(3756), 1583–1585.
• Kahneman, D. (1973). Attention and Effort. Prentice-Hall.
• Kahneman, D., Slovic, P., & Tversky, A. (Eds.). (1982). Judgment under Uncertainty: Heuristics and Biases. Cambridge University Press.
• Kahneman, D., Fredrickson, B. L., Schreiber, C. A., & Redelmeier, D. A. (1993). "When more pain is preferred to less: Adding a better end." Psychological Science, 4(6), 401–405.
• Redelmeier, D. A., & Kahneman, D. (1996). "Patients' memories of painful medical treatments." Pain, 66(1), 3–8.
• Kahneman, D., Wakker, P. P., & Sarin, R. (1997). "Back to Bentham? Explorations of experienced utility." Quarterly Journal of Economics, 112(2), 375–406.
• Kahneman, D., & Tversky, A. (Eds.). (2000). Choices, Values, and Frames. Cambridge University Press.
• Redelmeier, D. A., Katz, J., & Kahneman, D. (2003). "Memories of colonoscopy: a randomized trial." Pain, 104(1–2), 187–194.
• Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
• Kahneman, D., Sibony, O., & Sunstein, C. R. (2021). Noise: A Flaw in Human Judgment. Little, Brown.

8.3 Neoclassical Baseline, Axiomatics & Methodological Replications

• Allais, M. (1953). "Le comportement de l'homme rationnel devant le risque." Econometrica, 21(4), 503–546.
• Benartzi, S., & Thaler, R. H. (1995). "Myopic loss aversion and the equity premium puzzle." Quarterly Journal of Economics, 110(1), 73–92.
• Frederick, S. (2005). "Cognitive reflection and decision making." Journal of Economic Perspectives, 19(4), 25–42.
• Gal, D., & Rucker, D. D. (2018). "The loss of loss aversion." Journal of Consumer Psychology, 28(3), 497–516.
• Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1990). "Experimental tests of the endowment effect." Journal of Political Economy, 98(6), 1325–1348.
• Kahneman, D., & Klein, G. (2009). "Conditions for intuitive expertise." American Psychologist, 64(6), 515–526.
• Kahneman, D., & Killingsworth, M. A. (2023). "Income and emotional well-being: A conflict resolved." PNAS, 120(10), e2208661120.
• Klein, R. A., et al. (2018). "Many Labs 2." Advances in Methods and Practices in Psychological Science, 1(4), 446–490.
• Prelec, D. (1998). "The probability weighting function." Econometrica, 66(3), 497–527.
• Quiggin, J. (1982). "A theory of anticipated utility." Journal of Economic Behavior & Organization, 3(4), 323–343.
• Rabin, M. (2000). "Risk aversion and expected-utility theory: A calibration theorem." Econometrica, 68(5), 1281–1292.
• Savage, L. J. (1954). The Foundations of Statistics. John Wiley & Sons.
• Stanovich, K. E., & West, R. F. (2000). "Individual differences in reasoning." Behavioral and Brain Sciences, 23(5), 645–665.
• Von Neumann, J., & Morgenstern, O. (1944). Theory of Games and Economic Behavior. Princeton University Press.