• General Overview
  • Central Thesis: Two Systems of Thought
    • Two systems: System 1 (fast, automatic, intuitive) and System 2 (slow, deliberate, effortful)
    • System 1 is the secret author: most impressions and decisions arise without conscious awareness
    • System 2 is lazy: endorses intuitive answers without scrutiny, rarely noticing it answered the wrong question
    • Cognitive ease: System 1's dial signaling safety; linked to good mood, intuition, and gullibility
    • WYSIATI: System 1 constructs the best story from only activated ideas, ignoring missing information
  • Heuristics and Biases
    • Representativeness: judging probability by resemblance to a stereotype, ignoring base rates
    • Availability: judging frequency by ease of recall; vivid, dramatic events are overestimated
    • Anchoring: estimates stay close to a considered value, even a random one
    • Conjunction fallacy: judging a conjunction as more probable than its constituent (Linda problem)
    • Causal vs. statistical thinking: System 1 thinks causally, not statistically; base rates are ignored
  • Overconfidence and Illusions
    • Illusion of understanding: we construct simple stories of the past that overemphasize talent and ignore luck
    • Hindsight bias: once we know the outcome, we reconstruct past beliefs to match it
    • Illusion of validity: subjective confidence reflects story coherence, not evidence quality
    • Planning fallacy: forecasts unrealistically close to best-case scenarios, ignoring base rates
    • Optimistic bias: entrepreneurs and CEOs overestimate benefits and underestimate costs
  • Expert Intuition and Formulas
    • Intuition as recognition: Herbert Simon defined it as nothing more than stored pattern recognition
    • Two conditions for skill: a sufficiently regular environment and prolonged practice
    • Algorithms beat experts: ~60% of 200+ studies show statistical formulas more accurate than clinical judgment
    • Simple formulas work: equal-weight models often match optimal regression, unaffected by sampling accidents
    • The "broken-leg rule": formulas can be overruled only for rare, decisive exceptions
  • Prospect Theory and Loss Aversion
    • Reference dependence: people evaluate outcomes as gains or losses relative to a status quo
    • Loss aversion: losses loom larger than gains, typically 1.5–2.5 times more impactful
    • Diminishing sensitivity: concave for gains (risk averse), convex for losses (risk seeking)
    • Fourfold pattern: risk preferences flip depending on probability (high/low) and domain (gain/loss)
    • Framing effects: logically equivalent statements evoke opposite preferences (Asian disease problem)
  • The Remembering Self vs. The Experiencing Self
    • Peak-end rule: global memory of an experience is predicted by its worst moment and its end
    • Duration neglect: the length of a painful procedure has no effect on retrospective ratings
    • Two selves conflict: the remembering self governs decisions, not the actual experience
    • Focusing illusion: "Nothing in life is as important as you think it is when you are thinking about it"
    • Miswanting: bad choices from errors of affective forecasting, driven by the focusing illusion
  • Practical Remedies
    • Reference class forecasting: identify a reference class, obtain its statistics, generate a baseline prediction
    • Premortem: before final commitment, imagine the plan has failed and write its history
    • Risk policies: pre-set rules embed each decision in a statistical aggregate, countering optimism and loss aversion
    • Choice architecture: nudges (defaults, framing) preserve freedom while steering toward better choices
    • Organizational discipline: checklists, independent judgments, and orderly procedures improve decisions
• Deep Dive
  • Introduction
    • The Watercooler Goal
      • Gossip as tool: easier to spot others' mistakes than our own; informed gossip fuels self-criticism
      • Richer vocabulary: systematic errors (biases) need diagnostic labels like "halo effect" to be recognized
      • Silent mind: most impressions and decisions arise without conscious awareness of their origin
      • Focus on error: does not denigrate intelligence, just as medical texts on disease don't deny health
    • Origins: The Collaboration
      • 1969 meeting: Kahneman asked Tversky to guest-lecture; they debated intuitive statistics and found it fun
      • First study: even expert statisticians overestimated replicability from small samples—intuition was biased
      • Shared mind: they spent years in continuous conversation, checking criticism at the door, enjoying perfectionism
      • Heuristics discovered: resemblance (representativeness) and ease of recall (availability) produced predictable biases
      • 1974 Science article: "Judgment Under Uncertainty" documented 20+ biases, challenging the assumption of human rationality
    • Where We Are Now
      • Expert intuition: not magic but recognition—thousands of hours of practice let experts see patterns instantly
      • Affect heuristic: when a hard question is answered by an easier one (e.g., "Do I like Ford cars?" instead of "Is Ford stock underpriced?")
      • Two systems: System 1 (fast, automatic) and System 2 (slow, deliberate); System 1 is the secret author of most choices
      • Broader picture: skill and heuristics are both sources of intuitive judgment; emotion now looms larger than in early work
    • What Comes Next
      • Part 1: two-systems model, associative memory, and the automatic processes behind heuristics
      • Part 2: why statistical thinking is hard—System 1 thinks causally, not statistically
      • Part 3: overconfidence, illusion of certainty, and the role of chance (influenced by Taleb's The Black Swan)
      • Part 4: prospect theory, framing effects, and the challenge to rational-agent economics
      • Part 5: experiencing self vs. remembering self; implications for well-being and organizational decision-making
  • Part 1: TWO SYSTEMS
    • Chapters 1–2
      • The Two Systems
        • System 1: operates automatically, quickly, with little effort and no voluntary control
        • System 2: allocates attention to effortful mental activities, including complex computations
        • System 1 generates impressions and feelings; System 2 turns them into beliefs and deliberate choices
        • System 1 runs continuously; System 2 normally operates in a low-effort mode
        • When System 1 hits difficulty, it calls on System 2 for detailed processing
        • Conflict: System 2 must often override automatic impulses of System 1 (self-control)
      • Attention and Mental Effort
        • System 2 operations require attention and are disrupted when attention is diverted
        • Pupil dilation: a reliable measure of mental effort — dilates up to 50% during hard tasks like Add-3
        • Law of least effort: people gravitate to the least demanding course of action; laziness is built in
        • Intense focus can cause "inattentional blindness" — missing a gorilla crossing a basketball court
        • The mind normally operates at a comfortable walk, with rare sprints of intense effort
      • Capabilities and Limitations
        • System 1 excels at detecting simple relations and integrating information about one thing
        • System 2 can follow rules, compare multiple attributes, and make deliberate choices
        • Only System 2 can adopt "task sets" — programming memory to override habitual responses
        • Switching tasks is effortful, especially under time pressure
        • As skill increases, mental energy demand decreases; talent has similar effects
      • Cognitive Illusions
        • Müller-Lyer illusion: knowing the lines are equal doesn't stop System 1 from seeing them as different
        • Cognitive illusions: systematic errors that System 1 is prone to make in specific circumstances
        • System 1 cannot be turned off; biases are difficult to prevent
        • Best defense: learn to recognize situations where mistakes are likely, especially when stakes are high
        • It is easier to recognize others' mistakes than our own
    • 3: The Lazy Controller
      • The Effort of Walking and Thinking
        • Dual-task competition: walking and thinking compete for System 2's limited resources; hard calculations make you stop in your tracks
        • Law of least effort: people avoid effortful mental work when possible, as it requires self-control
        • Flow state: effortless deep concentration on absorbing tasks requires no self-control, freeing resources for the task
      • The Busy and Depleted System 2
        • Cognitive load weakens control: busy System 2 makes you more likely to yield to temptation, make selfish choices, and rely on superficial judgments
        • Ego depletion: exerting self-control in one task reduces willingness to exert it in the next; all voluntary effort draws on a shared pool
        • Glucose as fuel: effortful mental activity depletes blood glucose; restoring it can undo ego-depletion effects
        • Parole judge study: tired and hungry judges default to denying parole, with approval rates dropping to near zero before meals
      • The Lazy System 2
        • Bat-and-ball puzzle: intuitive answer (10¢) is wrong; over 50% at elite universities fail to check it, revealing lazy monitoring
        • Flowers syllogism: most endorse a flawed argument because a plausible conclusion comes first and feels true
        • Michigan/Detroit problem: failing to retrieve relevant facts from memory reflects both System 1's memory function and System 2's willingness to search
        • Intellectual sloth: failing these minitests is largely a matter of insufficient motivation, not inability
      • Intelligence, Control, and Rationality
        • Mischel's marshmallow test: children who resisted temptation at age 4 had higher intelligence and executive control as teenagers
        • Attention training: computer games that improve attention control also raise nonverbal intelligence scores
        • Cognitive Reflection Test: low scorers are impulsive, impatient, and prefer immediate gratification over larger delayed rewards
        • Rationality vs. intelligence: Stanovich argues that high IQ does not prevent bias; rationality—the reflective mind's willingness to override intuition—is a separate ability
    • Chapters 4–5
      • The Associative Machine
        • Associative activation: an idea triggers a spreading cascade of related ideas, emotions, and physical reactions, all automatically and outside conscious control
        • Associative coherence: the triggered elements (thoughts, feelings, bodily responses) reinforce each other into a self-consistent pattern
        • Cognition is embodied: you think with your body—facial expressions and gestures shape emotions and judgments, not just reflect them
        • Ideomotor effect: thinking of old age makes you walk slowly; walking slowly makes you think of old age—reciprocal priming between ideas and actions
        • Priming reaches everywhere: subtle cues (money images, school settings, eyes on a poster) unconsciously influence behavior, from selfishness to honesty
      • Cognitive Ease and Strain
        • Cognitive ease: a System 1 dial signaling that things are going well—no threats, no need for effort; linked to good mood, intuition, and gullibility
        • Cognitive strain: signals a problem, mobilizes System 2—leads to vigilance, analytic thinking, fewer errors, but less creativity
        • Illusions of truth and familiarity: repeated exposure or clear fonts make statements feel true; familiarity is easily mistaken for accuracy
        • Mere exposure effect: repeated safe stimuli become liked, even when presented too briefly for conscious awareness—a biological survival mechanism
        • Mood governs intuition: good mood doubles accuracy on intuitive coherence tasks; bad mood destroys intuitive performance entirely
    • Chapters 6–7
      • Norms, Surprises, and Causes
        • System 1's main function: maintains a model of what is normal in your world, built from associative links between recurring events.
        • Two varieties of surprise: active (waiting for a specific event) and passive (recognizing an event as normal without expecting it).
        • Norm theory: a single incident can alter what feels normal, making a recurrence less surprising even if objectively less probable.
        • Moses illusion: we accept "Moses took animals into the ark" because associative coherence (biblical context) overrides factual error.
        • Seeing causality: System 1 automatically constructs causal stories from events, as with the soup-wincing patron or the rising/falling bond prices on Hussein's capture.
        • Physical and intentional causality: we perceive causation directly (Michotte's launching squares) and infer intentions (Heider & Simmel's moving shapes), even in infancy.
        • Causal intuitions vs. statistical thinking: System 1 lacks the capacity for statistical reasoning, leading to inappropriate causal explanations.
      • A Machine for Jumping to Conclusions
        • Neglect of ambiguity: System 1 resolves ambiguous stimuli (e.g., "A 13 C" vs. "12 B 14") by betting on the most likely interpretation, without awareness of alternatives.
        • Bias to believe: System 1 automatically attempts to believe any statement; System 2 must actively doubt, but is often lazy or busy.
        • Confirmation bias: associative memory and System 2 both favor seeking evidence that confirms existing beliefs rather than refuting them.
        • Halo effect: liking one trait (e.g., intelligence) colors perception of unrelated traits (e.g., generosity), creating a coherent but spurious impression.
        • Decorrelation principle: to reduce error, gather independent judgments before discussion; open meetings amplify the first speaker's influence.
        • WYSIATI (What You See Is All There Is): System 1 constructs the best story from only activated ideas, ignoring missing information—leading to overconfidence, framing effects, and base-rate neglect.
    • Chapters 8–9
      • How Judgments Happen
        • System 1's continuous monitoring: generates basic assessments of situations (threat, opportunity, normality) without effort or intention
        • Basic assessments: friend/foe discrimination from facial features—dominance (chin shape) and trustworthiness (expression)
        • Voting heuristic: voters judge candidate competence from facial appearance in under a second, predicting election winners ~70% of the time
        • Mental shotgun: intention to answer one question triggers excess computations (e.g., spelling interferes with rhyme detection)
        • Prototype vs. sum: System 1 represents categories by prototypes, ignoring quantity—people donate same amount to save 2,000 or 200,000 birds
        • Intensity matching: System 1 translates across dimensions (e.g., matching crime severity to punishment loudness)
      • Answering an Easier Question
        • Substitution: when a hard target question stumps System 1, it answers an easier heuristic question instead
        • Heuristic question: simpler, emotionally charged alternative that comes readily to mind (e.g., "How angry do I feel?" for "What punishment fits?")
        • Mood heuristic for happiness: asking about dating first makes romantic satisfaction substitute for general happiness, creating spurious correlation
        • Affect heuristic: likes and dislikes determine beliefs about risks and benefits—System 2 becomes an apologist for System 1's emotions
        • 3-D heuristic: automatic substitution of 3-D size for 2-D size creates a powerful, unavoidable illusion (figures appear different sizes on the page)
        • Lazy System 2: endorses heuristic answers without scrutiny, rarely noticing it answered the wrong question
  • Part 2: HEURISTICS AND BIASES
    • 10: The Law of Small Numbers
      • The Illusion of Pattern in Small Samples
        • Law of Small Numbers: the mistaken belief that the law of large numbers applies to small samples too.
        • Extreme outcomes (high or low) are far more likely in small samples than large ones, purely by chance.
        • Rural counties show both the highest and lowest kidney cancer rates—not due to lifestyle, but to small population size.
        • No causal explanation is needed; the variation is a statistical artifact of sampling.
      • The Bias of Confidence Over Doubt
        • WYSIATI: System 1 focuses on the story, not the reliability of its source.
        • System 1 suppresses doubt and constructs coherent narratives from scant evidence.
        • System 2 can doubt, but sustaining uncertainty is harder work than sliding into certainty.
        • Overconfidence in small samples is a general bias favoring certainty over doubt.
      • Cause vs. Chance
        • Associative machinery seeks causal explanations, but statistical regularities demand a different view.
        • Random sequences (e.g., BBBGGG) feel non-random, yet all sequences of equal length are equally likely.
        • Pattern-seeking was evolutionarily advantageous, but leads to misclassifying random events as systematic.
        • Hot hand fallacy: streaks in basketball are cognitive illusions; shot sequences pass all tests of randomness.
      • Real-World Consequences
        • Gates Foundation invested $1.7B in small schools based on high performers, ignoring that low performers are also small.
        • Small schools are more variable, not better; large schools often produce better results.
        • Jumping to conclusions is safer in imagination than in reality; causal explanations of chance events are always wrong.
    • Chapters 11–12
      • Anchoring
        • Anchoring effect: estimates stay close to a considered value, even a random one (e.g., wheel of fortune at 10 or 65 shifted UN estimates by 20%)
        • Two mechanisms: deliberate System 2 adjustment, and automatic System 1 priming
        • Insufficient adjustment: a weak/lazy System 2 stops adjusting too early (e.g., driving too fast after exiting a highway)
        • Priming effect: the anchor activates compatible memories (e.g., high temperature primes "sun," low primes "frost"), biasing subsequent estimates
        • Anchoring index: a measurable ratio; typical values of 30–55% in experiments, even for experts like real-estate agents
        • Resistance: "thinking the opposite" and activating System 2 reduces anchoring; storming out of a negotiation can break the anchor's grip
      • The Science of Availability
        • Availability heuristic: judging frequency by the ease with which instances come to mind; substitutes a hard question (frequency) for an easy one (fluency)
        • Biases from availability: salient, dramatic, or personal events are overestimated (e.g., plane crashes, Hollywood divorces)
        • Fluency trumps count: listing 12 (hard) instances of assertiveness makes people rate themselves less assertive than listing 6 (easy) instances
        • Surprise drives the effect: the heuristic relies on unexpected difficulty; explaining fluency (e.g., "music makes retrieval hard") eliminates the bias
        • System 1 vs. System 2: people in good moods, under cognitive load, or feeling powerful rely more on fluency; engaged System 2 focuses on content instead
    • Chapters 13–14
      • 13: Availability, Emotion, and Risk
        • Availability and disasters: Concern spikes after a disaster, then fades as memories dim, creating cycles of panic and complacency.
        • Media distortion: Unusual, vivid events (e.g., botulism) are overestimated; common causes (e.g., asthma) are underestimated.
        • Affect heuristic: People answer "How do I feel about it?" instead of "What do I think about it?" — emotions guide risk/benefit judgments.
        • Associative coherence: If you like a technology, you see high benefits and low risks; dislike flips the pattern, avoiding tradeoffs.
        • Experts vs. public: Experts count lives lost; the public distinguishes "good" vs. "bad" deaths — a richer, value-laden view of risk.
        • Availability cascade: A minor event, amplified by media and emotion, triggers public panic and government overreaction (e.g., Love Canal, Alar scare).
        • Probability neglect: We focus on the vivid numerator (the tragic story) and ignore the denominator, overreacting to tiny risks.
        • Terrorism: Exploits availability — gruesome images dominate, even though traffic deaths are far more common.
      • 14: Tom W’s Specialty
        • Base-rate neglect: Given a personality sketch, people ignore the size of each field and predict by similarity to a stereotype.
        • Representativeness heuristic: Judging probability by how well an individual matches a stereotype — easy, intuitive, but often wrong.
        • Sins of representativeness: Overpredicting rare events (low base rates) and ignoring the quality or trustworthiness of evidence.
        • WYSIATI: Your mind treats available information as true, even when told it may be unreliable.
        • Bayesian discipline: Anchor on base rates; question the diagnosticity of your evidence — both are straightforward but rarely practiced.
        • Frowning helps: Activating System 2 (e.g., by frowning) reduces reliance on representativeness and improves use of base rates.
    • Chapters 15–16
      • 15: Linda: Less is More
        • Conjunction Fallacy: judging a conjunction (feminist bank teller) as more probable than its constituent (bank teller), violating logic.
        • Representativeness vs. Probability: System 1 substitutes plausibility and coherence for probability, making detailed scenarios seem more likely.
        • Less-is-More Pattern: adding broken dishes to a set lowers its perceived value in single evaluation; System 1 averages instead of adding.
        • Joint vs. Single Evaluation: direct comparison often corrects the fallacy, but Linda's case resists correction because plausibility overwhelms logic.
        • Frequency Format: phrasing "how many out of 100?" activates spatial reasoning, reducing fallacy rates from 65% to 25%.
        • System 2 Laziness: most participants "knew" the logic but did not apply it, content to answer as if asked for an opinion.
      • 16: Causes Trump Statistics
        • Statistical vs. Causal Base Rates: people ignore statistical base rates (85% Green cabs) but use causal ones (Green cabs cause 85% of accidents) to form stereotypes.
        • Causal Stereotypes: System 1 treats group statistics as causal traits of individuals, improving Bayesian reasoning but risking harmful profiling.
        • Resistance to Learning: students told surprising statistical facts (only 27% helped a seizure victim) did not change their predictions about individuals.
        • Particular Over General: people readily infer the general from a surprising individual case, but not the particular from general statistics.
        • Teaching Psychology: to change beliefs, you must surprise with individual cases, not mere statistical facts.
    • Chapters 17–18
      • Regression to the Mean
        • The flight instructor’s fallacy: praise follows exceptional performance, which then regresses; punishment follows poor performance, which then improves. The causal story is wrong.
        • Success = talent + luck: great success = a little more talent + a lot of luck. Extreme outcomes are partly lucky, so they regress.
        • The Sports Illustrated jinx: cover athletes perform worse next season. No curse needed—just regression from a lucky peak.
        • Correlation and regression are one concept: whenever correlation is imperfect, regression to the mean is mathematically inevitable.
        • Causal stories are invented: we explain regression (e.g., “relaxed after a bad jump”) when the true cause is just statistical noise.
        • Depressed children improve with any treatment: extreme groups regress naturally; a control group is essential to prove causation.
      • Taming Intuitive Predictions
        • Intuitive predictions substitute evaluation for forecasting: people match the extremeness of the prediction to the extremeness of the evidence, ignoring uncertainty.
        • Nonregressive predictions are biased: they are too optimistic for those who did well and too pessimistic for those who did poorly.
        • The correction formula: start with the baseline average, then move only partway toward your intuitive prediction, proportional to the correlation strength.
        • Unbiased predictions never call extreme cases: this is correct but unsatisfying; we prefer the thrill of saying “I thought so!”
        • Venture capitalists may need extreme language: if missing a rare success is far worse than overestimating many failures, biased predictions can be strategic.
        • Small samples produce more extreme impressions: a brilliant talk with no track record (Kim) should be regressed more than a solid record with a less dazzling presentation (Jane).
  • Part 3: OVERCONFIDENCE
    • 19: The Illusion of Understanding
      • The Narrative Fallacy
        • Narrative fallacy: we construct simple, concrete stories of the past that overemphasize talent and intentions while ignoring luck.
        • Halo effect: exaggerates consistency—good people do only good things, bad people only bad—making stories more coherent.
        • Illusion of inevitability: a compelling story makes the actual outcome seem like the only possible one.
        • WYSIATI: we build the best story from limited information and believe it, ignoring what we don't know.
      • Hindsight Bias and Its Costs
        • I-knew-it-all-along effect: once we know the outcome, we reconstruct our past beliefs to match it, underestimating surprise.
        • Outcome bias: we judge decisions by their results, not by the quality of the process at the time.
        • Pernicious effect: hindsight punishes prudent risk-takers whose good decisions happen to fail, and rewards reckless gamblers who get lucky.
        • Bureaucratic defensiveness: fear of hindsight scrutiny leads to excessive risk aversion and defensive procedures.
      • Illusions of Success in Business
        • CEO impact is small: correlation between CEO quality and firm success is about .30—only a 10% edge over random guessing.
        • Halo effect in reverse: a successful CEO is called flexible; the same person, after failure, is called rigid.
        • Regression to the mean: the gap between successful and less successful firms shrinks over time because luck, not skill, drove much of the original difference.
        • Business books exploit demand: stories of success and failure offer illusory certainty, but their lessons have little enduring value.
    • 20: The Illusion of Validity
      • The Illusion of Validity in Officer Selection
        • System 1's leap: jumps to conclusions from little evidence, unaware of the jump's size.
        • Confidence by coherence: subjective confidence reflects the story's coherence, not evidence quality.
        • The army test: watching soldiers on an obstacle field produced compelling but useless predictions of leadership.
        • Nonregressive predictions: strong forecasts of success or failure were made from weak evidence (WYSIATI).
        • Ignoring base rates: knowing general predictions were worthless did not shake confidence in individual cases.
        • Confidence is a feeling: it signals a coherent story, not a valid judgment.
      • The Illusion of Stock-Picking Skill
        • The puzzle of trading: buyers and sellers both believe the current price is wrong, an illusion for most.
        • Individual investors lose: those who trade most earn the least; selling winners and buying losers hurts returns.
        • Professionals fail the skill test: year-to-year performance correlations among fund managers are near zero.
        • Luck rewarded as skill: firms compensate traders based on outcomes that are essentially random.
        • Culture of denial: statistical evidence challenging the illusion of skill is quickly ignored or forgotten.
      • The Illusions of Pundits
        • Hindsight bias: coherent narratives of the past make the future seem more predictable than it is.
        • Tetlock's study: expert political and economic forecasts were worse than random chance.
        • Hedgehogs vs. foxes: hedgehogs (one big theory) are overconfident and poor predictors; foxes (complex thinkers) are slightly better.
        • Overconfidence increases with knowledge: more expertise often leads to a stronger illusion of skill.
      • The Unpredictable World
        • Errors are inevitable: the world is fundamentally unpredictable, especially in the long term.
        • Trust low confidence more: high subjective confidence is a poor indicator of accuracy.
        • Limited predictability: short-term trends and past behavior can predict future behavior, but with modest validity.
        • The key question: not whether experts are trained, but whether their world is predictable.
    • 21: Intuitions vs. Formulas
      • The Meehl Pattern: Algorithms Beat Experts
        • Paul Meehl's "disturbing little book": reviewed 20 studies where statistical formulas outperformed clinical predictions
        • Consistent results: ~60% of 200+ studies show algorithms significantly more accurate; the rest tie, which is a win for cheaper formulas
        • No documented exception: experts matched or exceeded by simple algorithms across medicine, economics, parole, and wine pricing
        • Orley Ashenfelter's wine formula: predicts Bordeaux prices from weather data (summer temp, harvest rain, winter rain) with >0.90 correlation, beating expert tasters
      • Why Experts Fail
        • Overcomplexity: experts try to be clever with complex combinations; simple combinations of features are more valid
        • Inconsistency: humans contradict themselves ~20% of the time on identical cases (e.g., radiologists reading X-rays)
        • Context dependency of System 1: unnoticed stimuli (e.g., a cool breeze, judge's food breaks) fluctuate judgments moment to moment
        • The "broken-leg rule": formulas can be overruled only for rare, decisive exceptions; otherwise, substituting judgment harms accuracy
      • The Power of Simple Formulas
        • Robyn Dawes's "improper linear models": equal-weight formulas often match or beat optimal multiple regression, unaffected by sampling accidents
        • Marital stability formula: frequency of lovemaking minus frequency of quarrels — a back-of-envelope algorithm that works
        • Virginia Apgar's test: five variables (heart rate, respiration, reflex, muscle tone, color) scored 0–2 each; saved hundreds of thousands of infant lives
        • Checklists and simple rules: Atul Gawande's A Checklist Manifesto extends this principle to surgery and other high-stakes domains
      • Hostility to Algorithms
        • Clinicians' illusion of skill: short-term hunches in therapy are confirmed, but long-term predictions (where formulas excel) are never learned properly
        • Moral preference for human error: a child dying from an algorithm mistake feels more tragic than the same from human error, though both are equally fatal
        • "All natural" bias: preference for the organic, the human, the intuitive — even when the artificial is proven superior
        • Gradual acceptance: as algorithms recommend books, set credit limits, and guide sports decisions, resistance softens
      • Learning from Meehl: A Personal Case
        • Kahneman's army interview redesign (1955): replaced global intuitive judgments with six specific traits scored separately via factual questions
        • Result: the sum of six ratings was a substantial improvement over the old interview, though still imperfect
        • Surprise finding: the "close your eyes" intuitive judgment after disciplined scoring did just as well as the formula
        • Lesson: do not trust intuition alone, but do not dismiss it — use it only after disciplined collection of objective information
      • Practical Application: Hiring
        • Procedure: select 3–6 independent traits, prepare factual questions and scoring scales for each, score one trait at a time to avoid halo effects
        • Final decision: add up the six scores and hire the candidate with the highest total — resist inventing "broken legs" to change the ranking
        • Promise: this disciplined method is far more likely to find the best candidate than unprepared intuitive interviews
    • 22: Expert Intuition: When Can We Trust It?
      • The Nature of Intuitive Expertise
        • Intuition as recognition: Herbert Simon defined intuition as nothing more than recognition, where a cue gives access to stored memory.
        • Recognition-primed decision model: Experts generate a single plausible option via System 1, then simulate it mentally via System 2.
        • Subjective confidence is unreliable: Confidence stems from cognitive ease and coherence, not validity; do not trust anyone’s self-assessment of their judgment.
        • Skill acquisition requires two conditions: a sufficiently regular environment and prolonged practice to learn its regularities.
      • Environments That Support or Undermine Skill
        • High-validity environments: Chess, firefighting, anesthesiology—stable regularities allow genuine intuitive expertise.
        • Low-validity environments: Long-term stock picking, political forecasting—unpredictable; intuitive “hits” are luck or lies.
        • Wicked environments: Professionals learn wrong lessons (e.g., early physician spreading typhoid by unwashed hands).
        • Statistical algorithms excel in noise: They detect weak cues and apply them consistently, outperforming humans in low-validity settings.
      • The Critical Role of Feedback and Practice
        • Immediate feedback enables skill: Braking a car offers instant reward/punishment; harbor pilots struggle due to delayed outcomes.
        • Expertise is task-specific: A psychotherapist may read a patient’s mood well but cannot forecast long-term outcomes; radiologists get sparse feedback on missed diagnoses.
        • Overconfidence arises from unrecognized limits: Professionals often mistake skill in one task for skill in another (e.g., short-term anticipation vs. long-term forecasting).
      • Distinguishing Valid from Bogus Intuition
        • Evaluate provenance, not the intuition itself: Check if the environment is regular and the expert had adequate learning opportunity.
        • Substitution creates false coherence: System 1 answers an easier question, producing plausible but wrong intuitions that System 2 lazily endorses.
        • The Klein-Kahneman agreement: Despite emotional differences, they agreed that regularity of environment and learning history, not confidence, determine trustworthiness.
    • 23: The Outside View
      • The Inside View vs. The Outside View
        • Inside view: forecast based on specific circumstances and plans, ignoring broader data
        • Outside view: baseline prediction derived from statistics of similar cases (reference class)
        • Key insight: the inside view is natural but often delusional; the outside view is more accurate but routinely ignored
        • Seymour's dual judgment: his inside estimate (2 years) contradicted his own knowledge of similar projects (7-10 years, 40% failure)
      • The Planning Fallacy
        • Planning fallacy: forecasts unrealistically close to best-case scenarios, ignoring base rates
        • Root cause: WYSIATI—we extrapolate from current progress and fail to anticipate "unknown unknowns"
        • Ubiquity: kitchen renovations (100% cost overrun), rail projects (106% passenger overestimate, 45% cost overrun), Scottish Parliament (£40M → £431M)
        • Self-serving bias: planners may deliberately underestimate to gain approval, knowing projects rarely get abandoned
      • Mitigating the Planning Fallacy
        • Reference class forecasting: identify a reference class, obtain its statistics, generate a baseline prediction, then adjust for case specifics
        • Key advice (Flyvbjerg): "The prevalent tendency to underweight distributional information is perhaps the major source of error in forecasting"
        • Organizational remedy: reward precise execution; penalize failure to anticipate difficulties and unknown unknowns
      • Decisions and Errors
        • Delusional optimism: executives overestimate benefits, underestimate costs, and pursue risky initiatives based on unrealistic odds
        • Consequences: explains why people litigate, start wars, and open small businesses despite poor odds
        • Sunk-cost fallacy: reluctance to abandon a failing project due to prior investment and embarrassment
        • Author's confession: despite knowing the outside view, the team irrationally persevered for 8 years on a project that was never used
    • 24: The Engine of Capitalism
      • The Optimistic Bias
        • Optimistic bias: pervasive tendency to view the world as more benign, ourselves as more favorable, and goals as more achievable than reality
        • Inherited disposition: optimism is largely genetic, part of a general well-being trait that prefers seeing the bright side
        • Blessings of mild optimism: cheerfulness, resilience, better health, longer life, and greater persistence
        • Disproportionate influence: optimistic individuals are the inventors, entrepreneurs, and leaders who shape our lives by seeking challenges and taking risks
      • Entrepreneurial Delusions
        • Statistical blindness: 81% of entrepreneurs rate their success odds at 7/10 or higher, though only 35% of small businesses survive five years
        • Costly persistence: 47% of inventors double their losses after receiving objective failure predictions, especially those high in optimism
        • Mediocre returns: self-employment yields lower average returns than selling skills to employers
        • Hubris hypothesis: overconfident CEOs overpay for acquisitions and destroy shareholder value, even when they personally have more at stake
      • Competition Neglect
        • Cognitive root: WYSIATI makes entrepreneurs focus on their own plans while neglecting competitors' skills and plans
        • Above-average effect: people answer "Am I better than average?" by substituting the easier question "Am I good at this?"
        • Excess entry: more competitors enter markets than can profitably sustain them, producing average losses for entrants
        • Optimistic martyrs: failed entrepreneurs signal new markets to more qualified competitors, benefiting the economy but harming investors
      • Overconfidence
        • Worthless forecasts: CFOs' S&P return predictions correlate slightly below zero with actual outcomes, yet they remain unaware
        • 80% confidence intervals: CFOs' stated intervals produce 67% surprises (expected 20%), revealing gross overconfidence
        • Social pressure: experts who admit ignorance are replaced by more confident competitors who gain client trust
        • Bold forecasts, timid decisions: risk takers are not thrill-seekers but simply less aware of risks than cautious people
      • The Premortem: A Partial Remedy
        • Procedure: before final commitment, imagine the plan has failed disastrously and write a brief history of that failure
        • Legitimizes doubt: overcomes groupthink by encouraging supporters to search for unconsidered threats
        • Limited cure: reduces damage from WYSIATI and uncritical optimism but does not provide complete protection
  • Part 4: CHOICES
    • Chapters 25–26
      • Bernoulli’s Errors
        • Expected utility theory: a logic of rational choice, not a psychological model
        • Bernoulli’s insight: diminishing marginal utility of wealth explains risk aversion
        • Theory-induced blindness: scholars ignored obvious counterexamples for 250 years
        • Missing reference point: Jack and Jill have same wealth but opposite happiness due to recent changes
        • Anthony vs. Betty: same gamble evaluated as gains or losses depending on current wealth
        • Flaw: utility depends on changes from a reference point, not absolute states of wealth
      • Prospect Theory
        • Outcomes as gains/losses: reference point replaces wealth as carrier of value
        • Three cognitive features: reference dependence, diminishing sensitivity, loss aversion
        • Loss aversion: losses loom larger than gains, typically 1.5–2.5 times more impactful
        • Risk seeking for losses: sure loss is so aversive people prefer a gamble with worse expected value
        • Mixed gambles: loss aversion causes extreme risk aversion; bad options cause risk seeking
        • Rabin’s proof: rejecting small favorable gambles implies absurd risk aversion for large ones
      • Blind Spots of Prospect Theory
        • Disappointment: winning nothing feels like a large loss when a big win was almost certain
        • Regret: outcomes depend on foregone alternatives, not just the chosen option
        • Simplicity wins: regret theories make few striking predictions beyond prospect theory
        • Acceptance criterion: new concepts (reference point, loss aversion) earned their place by yielding true predictions
    • Chapters 27–28
      • The Endowment Effect
        • Reference point: The status quo is omitted from standard indifference maps, causing "theory-induced blindness"
        • Loss aversion: Disadvantages of change loom larger than advantages, creating a bias toward the status quo
        • Tastes are not fixed: Preferences vary with the reference point; what you own is valued more than what you don't
        • Held for use vs. exchange: Goods intended for consumption trigger loss aversion; routine trade goods do not
        • Mugs experiment: Sellers demanded ~$7, while buyers offered ~$3 — a 2:1 ratio reflecting loss aversion
        • Thinking like a trader: Experienced traders overcome the endowment effect by focusing on opportunity costs
      • Bad Events
        • Negativity dominance: The brain prioritizes threats over opportunities; bad news is processed faster and more thoroughly
        • Goals as reference points: Not achieving a goal is a loss; the aversion to failure is stronger than the desire to exceed
        • Golf putts study: Professionals putt more accurately for par (avoiding bogey) than for birdie, by 3.6%
        • Defending the status quo: Loss aversion makes negotiations and reforms difficult; defenders fight harder than challengers
        • Fairness and entitlements: It is unfair to impose losses on others to increase profit, but acceptable to pass on unavoidable losses
        • Legal asymmetry: Actual losses are compensated more than foregone gains; possession is nine-tenths of the law
    • 29: The Fourfold Pattern
      • Decision Weights vs. Probabilities
        • Expectation principle: values weighted by probability—poor psychology, not how we feel
        • Possibility effect: tiny chances (0→5%) are overweighted, creating hope disproportionate to odds
        • Certainty effect: near-certainty (95→100%) is underweighted; people pay heavily for absolute safety
        • Allais’s paradox: choices violate rational axioms because certainty effect distorts preferences
        • Measured weights: 2% chance gets weight 8.1 (4× over); 98% chance gets weight 87.1 (13% discount)
        • Extreme rarity: very rare events (<1%) are either ignored or massively overweighted when attention is drawn
      • The Fourfold Pattern of Risk Preferences
        • Top left (high prob gain): risk averse—sure gain preferred; fear of regret drives caution
        • Top right (high prob loss): risk seeking—desperate gambles to avoid sure loss; turns failures into disasters
        • Bottom left (low prob gain): risk seeking—lotteries thrive on possibility effect; people buy the right to dream
        • Bottom right (low prob loss): risk averse—insurance purchased for peace of mind, far above expected value
      • Applications in Legal Negotiation
        • Strong plaintiff (top left): risk averse, settles for less than expected value—defendant holds stronger hand
        • Weak defendant (top right): risk seeking, fights on despite high loss probability
        • Frivolous plaintiff (bottom left): overweights small win chance, bargains aggressively
        • Frivolous defendant (bottom right): overweights small loss chance, settles generously to buy safety
        • Long-run cost: systematic deviations from expected value—whether risk averse or seeking—are expensive
    • Chapters 30–31
      • Rare Events
        • Availability cascade: vivid, media-amplified images of rare dangers (e.g., bus bombings) trigger automatic fear, overriding rational probability estimates
        • Emotion trumps probability: the mere possibility of a vivid outcome (terror, lottery win) drives decisions; exact odds are irrelevant to System 1
        • Overestimation vs. overweighting: people both overestimate rare-event probabilities and assign excessive decision weight to them in choices
        • Confirmatory bias: focusing on a specific outcome (e.g., a team winning) primes memory for supporting scenarios, inflating its perceived likelihood
        • Denominator neglect: "1 death in 1,000" feels more threatening than "0.1% risk" because the vivid numerator (the victim) overshadows the diffuse denominator
        • Choice from experience: when rare events are not explicitly described and must be learned from repeated trials, they are underweighted, not overweighted
      • Risk Policies
        • Narrow vs. broad framing: considering each risky choice in isolation (narrow) leads to inconsistent, often inferior preferences; aggregating decisions (broad) reveals dominant options
        • Samuelson's paradox: rejecting a single favorable gamble (loss aversion) while accepting many identical gambles is logically inconsistent but psychologically natural
        • Loss aversion + narrow framing = costly: frequent checking of short-term outcomes amplifies pain of small losses, reducing risk-taking and wealth
        • "You win a few, you lose a few": the mantra enables broad framing, blunting emotional reactions to individual losses and allowing rational acceptance of favorable risks
        • Risk policy as outside view: a pre-set rule (e.g., "always take the highest deductible") embeds each decision in a statistical aggregate, countering both optimism and loss aversion
    • 32: Keeping Score
      • Mental Accounts
        • Mental accounting: narrow framing that organizes life into separate budgets, enabling control by a finite mind
        • Econs use comprehensive views; Humans create distinct accounts for spending, savings, and self-control
        • Sunk-cost fallacy: throwing good money after bad to avoid closing a losing mental account
        • Driving into a blizzard for a paid ticket exemplifies the error; the cost is already sunk
        • The fallacy traps people in poor jobs, unhappy marriages, and doomed projects
        • Training in economics and business reduces susceptibility to the sunk-cost fallacy
      • The Disposition Effect
        • Disposition effect: preference for selling winners over losers to close each mental account as a gain
        • Rational agents sell the stock least likely to perform well, ignoring purchase price
        • Selling losers offers tax advantages and future gains; selling winners costs 3.4% annually
        • The bias is a costly form of narrow framing that experienced investors overcome with System 2
      • Regret and Normality
        • Regret: counterfactual emotion triggered by easily imagined alternatives to reality
        • Abnormal actions (picking up a hitchhiker) produce more regret than habitual ones
        • Action bias: outcomes from action cause stronger regret than identical outcomes from inaction
        • Departures from the default option are natural candidates for regret and blame
        • Anticipated regret favors conventional, risk-averse choices in consumers and fund managers
      • Responsibility and Taboo Tradeoffs
        • Loss aversion intensifies when you are responsible for a bad outcome, raising selling prices far above buying prices
        • Parents reject trading child safety for money at any price, a taboo tradeoff
        • This emotional stance is incoherent: finite safety budgets are better spent on cost-effective protections
        • Precautionary principle: prohibits actions that might cause harm, paralyzing innovation like airplanes or vaccines
        • Enhanced loss aversion originates in System 1 and clashes with efficient risk management
      • Coping with Regret
        • Anticipate regret explicitly before deciding to reduce its sting when things go badly
        • Hindsight avoidance: be either very thorough or completely casual in long-term decisions
        • The psychological immune system makes actual regret less painful than anticipated
    • 33: Reversals
      • Single vs. Joint Evaluation
        • Preference reversal: judgments flip between single and joint evaluation
        • Single evaluation: System 1 emotional reactions dominate, using intensity matching
        • Joint evaluation: System 2 comparison activates broader principles, suppressing irrelevant features
        • Real-world problem: life mostly presents single evaluations, so WYSIATI leads to inconsistent moral intuitions
      • The Burglary Shooting
        • Scenario: victim shot in regular store vs. unfamiliar store
        • Joint evaluation: compensation should be identical—location is irrelevant to injury
        • Single evaluation: poignancy ("if only") drives much higher awards for the unfamiliar store
        • Principle vs. emotion: endorsed principles only govern when alternatives are explicitly compared
      • Challenging Economics
        • Lichtenstein & Slovic: people choose safe bet B but set higher selling price on risky bet A
        • Grether & Plott: economists tried to discredit the finding, but replicated it instead
        • Core violation: preferences depend on context, contradicting the rational-agent model's coherence doctrine
        • Impact: opened economics to psychological research despite resilient theoretical beliefs
      • Categories and Norms
        • Within-category: judgments are coherent (apples vs. peaches share a norm)
        • Across-categories: incoherence emerges (apples vs. steak have no shared norm)
        • Evaluability hypothesis: attributes like "number of entries" have no meaning in isolation, only in comparison
        • Dolphins vs. farmworkers: single evaluation favors dolphins (charming); joint evaluation favors humans (moral feature)
      • Unjust Reversals in Law
        • Mock jury study: burned child gets less than defrauded bank in single evaluation, more in joint
        • Legal flaw: jurors are prohibited from considering other cases, forcing single evaluation
        • Agency penalties: fines are coherent within agencies (OSHA vs. EPA) but absurd globally ($7,000 vs. $25,000)
        • Broader frames: serve rationality; narrow frames let System 1 emotional reactions dominate
    • 34: Frames and Reality
      • Emotional Framing
        • Framing effect: logically equivalent statements evoke different reactions because System 1 is not reality-bound
        • Costs vs. Losses: a bad outcome is more acceptable framed as a cost than as a loss
        • Brain evidence: emotional words (KEEP/LOSE) activate the amygdala, driving frame-consistent choices
        • Rationality index: subjects least susceptible to framing showed enhanced frontal activity, not more conflict
        • Medical framing: physicians chose surgery at 84% when told "90% survival" vs. 50% when told "10% mortality"
      • Empty Intuitions
        • Asian disease problem: risk-averse for gains (save 200 for sure), risk-seeking for losses (gamble to avoid 400 deaths)
        • No underlying preference: moral intuitions attach to frames, not to reality itself
        • Tax exemption paradox: favoring the poor via child exemptions logically requires favoring them via childless surcharges
        • System 2 dumbfounded: when inconsistency is revealed, people have no moral intuitions about the real problem
      • Good Frames
        • Lost tickets vs. lost cash: broader frames (general revenue) lead to more rational decisions than narrow mental accounts
        • MPG illusion: gallons-per-mile is a superior frame to miles-per-gallon for comparing fuel savings
        • Organ donation defaults: opt-out yields ~100% donation; opt-in yields ~4%—driven by System 2 laziness
        • Policy implication: skeptics of rational-agent theory are sensitive to how inconsequential factors determine preferences
  • Part 5: TWO SELVES
    • Chapters 35–37
      • Two Selves
        • Experienced utility: Bentham's measure of pain and pleasure moment-to-moment
        • Decision utility: economists' "wantability" guiding rational choice
        • Injection puzzle: paying more to reduce 6 to 4 injections than 20 to 18 is absurd if pain is identical
        • Peak-end rule: global memory of an experience is predicted by the average of its worst moment and its end
        • Duration neglect: the length of a painful procedure had no effect on patients' retrospective ratings
        • Tyranny of the remembering self: memories govern decisions, not the actual experience
      • Life as a Story
        • Narrative logic: stories care about peaks and endings, not duration
        • Jen's life: adding 5 slightly happy years to a very happy life lowered its judged desirability
        • Amnesic vacation: if memories were erased, the value of the experience plummets
        • Choosing by memory: vacationers' intentions to repeat a trip were predicted by final evaluation, not daily diaries
        • The experiencing self is a stranger: people pity their suffering self no more than a stranger in pain
      • Experienced Well-Being
        • U-index: percentage of time spent in an unpleasant emotional state
        • Day Reconstruction Method: reliving yesterday in episodes yields valid measures of daily affect
        • Time use matters: commuting and work are high-U; socializing and sex are low-U
        • Attention is key: pleasure from eating is diluted when combined with other activities
        • Money and happiness: income beyond ~$75,000/year yields zero gain in experienced well-being, though life satisfaction keeps rising
        • Life satisfaction vs. experience: they are related but distinct; education boosts evaluation, not daily feelings
      • The U-Index as Policy Tool
        • U-Index: proportion of time an individual spends in an unpleasant emotional state
        • Policy objective: reduce human suffering by lowering society's U-index
        • Priority areas: tackling depression and extreme poverty yields greatest well-being gains
      • Time and Happiness Trade-offs
        • Time control: easiest route to increased happiness is reclaiming time for enjoyed activities
        • Income paradox: beyond satiation point, more money buys pleasures but erodes ability to savor simple joys
    • 38: Thinking About Life
      • The Marriage Satisfaction Puzzle
        • Affective forecasting error: people marry expecting lasting bliss, but the data show a steep decline after the event
        • Heuristic substitution: survey respondents answer "How satisfied are you with your life?" by substituting "How happy am I thinking about my marriage right now?"
        • Attention wanes: the salience of marriage fades with time, so the apparent happiness surge is a measurement artifact, not a real change in well-being
        • Experienced well-being unchanged: married women spend less time alone but also less time with friends, more time on chores—net effect on happiness is zero
      • The Focusing Illusion
        • Core principle: "Nothing in life is as important as you think it is when you are thinking about it"
        • WYSIATI in action: when evaluating life, System 1 substitutes a salient aspect (e.g., climate) for the whole, ignoring all other determinants
        • California paradox: Midwesterners and Californians share the mistaken belief that Californians are happier; actual life satisfaction is identical
        • Car example: you overestimate pleasure from your car because you answer "How much do I enjoy it when I think about it?"—which is rarely
      • Adaptation and Attention
        • Paraplegia evidence: after one year, those who know a paraplegic estimate 41% bad mood; those who don't estimate 68%—outsiders fail to forecast adaptation
        • Colostomy paradox: experience sampling shows no happiness difference from healthy people, yet patients would trade years of life to avoid the condition
        • Adaptation = withdrawal of attention: most long-term circumstances (good or bad) are "part-time states" we inhabit only when we think about them
        • Exceptions: chronic pain, loud noise, and severe depression resist adaptation because they forcibly command attention
      • Miswanting and Time Neglect
        • Miswanting: bad choices from errors of affective forecasting, often driven by the focusing illusion
        • Attention-durable goods win: a weekly book club retains attention value; a new car becomes invisible—yet we overvalue the car and undervalue the social commitment
        • Duration neglect: the remembering self judges episodes by peak and end, not by total time; transitions (winning a lottery, diagnosis) dominate forecasts
        • Hybrid well-being: we cannot ignore either experienced happiness or the goals people set—both selves must be considered
  • Conclusions
    • Two Selves
      • Remembering self vs. experiencing self: the self that keeps score often conflicts with the self that actually lives.
      • Duration neglect & peak-end rule: memory ignores how long an experience lasted, focusing only on its peak and end.
      • Flawed choices: people willingly repeat more painful experiences because the remembering self prefers a better memory.
      • Time matters: the experiencing self’s finite resource is time, but the remembering self treats duration as irrelevant.
      • Distorted reflection: memory’s biases (peaks, ends, hindsight) produce a warped view of actual experience.
      • Both selves matter: a complete theory of well-being must consider what people want and what they actually experience.
    • Econs and Humans
      • Rationality defined: for economists, it means internal consistency, not reasonableness — a standard Humans cannot meet.
      • Humans are not irrational: they fail the rational-agent model but are not impulsive or emotional; they need help, not condemnation.
      • Libertarian paternalism: nudges (e.g., default enrollment) preserve freedom while steering people toward better choices.
      • Choice architecture: framing and defaults exploit System 1’s quirks to improve decisions without coercion.
      • Protection from exploitation: Humans need safeguards against fine-print contracts that Econs would read and reject.
      • Broad political appeal: policies like Save More Tomorrow unite conservatives and liberals by improving outcomes without limiting freedom.
    • Two Systems
      • System 1 is the source of most good: automatic intuition handles routine challenges accurately when skill and feedback are present.
      • System 1 also causes errors: heuristics, WYSIATI, and associative coherence produce predictable biases and cognitive illusions.
      • System 2 is not a paragon: it is lazy, limited, and often endorses System 1’s flawed intuitions rather than correcting them.
      • Educating intuition is hard: recognizing minefields in others is easier than slowing down your own System 1.
      • Organizations can improve: checklists, premortems, and reference-class forecasting impose orderly procedures that individuals cannot.
      • A richer vocabulary helps: precise labels (e.g., “anchoring,” “narrow framing”) enable constructive criticism and better decision-making culture.
  • Appendix A: Judgment Under Uncertainty: Heuristics and Biases
    • The Representativeness Heuristic
      • Definition: probability judged by degree of resemblance to a stereotype, not by statistical logic
      • Base-rate neglect: prior probabilities are ignored when a representative description is present
      • Sample-size insensitivity: small and large samples are judged equally representative of a population
      • Misconceptions of chance: people expect short random sequences to be locally representative (gambler's fallacy)
      • Illusion of validity: high confidence in predictions based on a good match, even when evidence is unreliable
      • Regression neglect: failure to expect regression to the mean leads to spurious causal explanations (praise vs. punishment)
    • The Availability Heuristic
      • Definition: frequency or probability judged by ease of bringing instances to mind
      • Retrievability biases: famous, salient, or recent events appear more numerous than equally frequent but less vivid ones
      • Search-set biases: ease of searching by first letter inflates estimates of words starting with that letter
      • Imaginability biases: ease of constructing scenarios distorts probability estimates (committee-size problem)
      • Illusory correlation: strong associative bonds cause overestimation of co-occurrence frequency
    • Adjustment and Anchoring
      • Definition: estimates start from an initial value and are adjusted insufficiently
      • Insufficient adjustment: arbitrary starting points (wheel of fortune) produce biased final estimates
      • Conjunction overestimation: probability of a chain of events is overestimated because anchor is the high elementary probability
      • Disjunction underestimation: probability of at least one failure in a system is underestimated
      • Tight confidence intervals: experts produce overly narrow probability distributions because adjustments from best estimates are insufficient
    • The Gambler's Fallacy and Internal Consistency
      • Gambler's fallacy: belief that a tail becomes more likely after a run of heads
      • Internal consistency alone is insufficient for rational probability judgments
      • Judgments must align with the judge's total web of beliefs about the world
      • No simple formal procedure exists to test compatibility across all beliefs
      • Rational judge strives for compatibility, not just internal consistency
    • Summary of Heuristics and Biases
      • Three core heuristics: representativeness, availability, and anchoring-and-adjustment
      • Highly economical and usually effective in everyday judgments
      • Systematic errors are predictable consequences of these mental shortcuts
      • Understanding heuristics can improve real-world decisions under uncertainty
  • Appendix B: Choices, Values, and Frames
    • The Psychophysics of Value
      • Value function shape: concave for gains (risk averse), convex for losses (risk seeking), and steeper for losses than gains
      • Loss aversion: a loss of $X hurts about twice as much as a gain of $X pleases
      • Reference dependence: people evaluate outcomes as gains or losses relative to a status quo, not as final states of wealth
      • Risk seeking in losses: most prefer an 85% chance to lose $1,000 over a sure loss of $800, even though the gamble has worse expectation
    • The Psychophysics of Chance
      • Nonlinear weighting: people overweight sure things and very low probabilities, but underweight moderate and high probabilities
      • Pseudo-certainty effect: a gamble presented as a sure gain in a sequential game is preferred over an objectively identical simple gamble
      • Probabilistic insurance aversion: reducing risk by half is worth far less than half the premium, because going from p/2 to 0 has disproportionate impact
      • Lottery and insurance appeal: overweighting of long shots makes lotteries attractive; overweighting of tiny disaster risks makes insurance attractive
    • Framing and Invariance
      • Invariance failure: logically equivalent descriptions (e.g., "200 saved" vs. "400 die") produce opposite risk preferences
      • Dominance violation: concurrent decisions (sure gain + risky loss) can produce a dominated combination that subjects fail to detect
      • Frame dependence is inescapable: people cannot naturally recode outcomes into a canonical representation, especially for health or safety
      • Remedy: test preferences by deliberately reframing the same problem in multiple ways
    • Mental Accounting and Transactions
      • Topical accounts: people evaluate savings relative to the item's price, not in absolute terms—$5 off a $15 calculator is more compelling than $5 off a $125 jacket
      • Ticket vs. cash effect: losing a ticket is posted to the "play" account, making a second purchase feel like overpaying; losing cash is not
      • Cost-loss distinction: a $5 payment to enter a lottery is more acceptable than a $5 loss from the same gamble, even though they are identical
      • Endowment effect: owning an asset raises its value; selling prices far exceed buying prices because losses loom larger than gains
    • Decision Values vs. Experience Values
      • Mismatch: what people choose often fails to predict what they will actually enjoy (e.g., ordering too much food when hungry)
      • Framing shapes experience: labeling an expense as a "cost" vs. a "loss" can alter the hedonic experience of the outcome itself
      • Hedonic adaptation: the reference point shifts quickly, so objective improvements produce only short-lived pleasure
    • Key Contributions in Decision Research
      • Bernoulli (1738): Early formal theory of risk measurement, foundation for expected utility
      • Savage (1954): Established subjective expected utility as normative standard
      • Kahneman & Tversky (1979): Prospect theory as descriptive alternative to expected utility
      • Tversky & Kahneman (1981): Demonstrated how framing systematically alters choices
      • Thaler (1980, 1985): Mental accounting explains consumer behavior beyond rational models
    • Empirical Methods and Applications
      • McNeil et al. (1982): Framing effects in medical treatment preferences
      • Fischhoff et al. (1980): Labile values revealed through elicitation methods
      • Knetsch & Sinden (1984): Disparity between willingness to pay and compensation demanded
      • Slovic et al. (1982): Response mode and framing shape risk assessment outcomes
  • Acknowledgments
    • The Collaborative Process
      • Jason Zweig: urged the project forward and generously contributed editorial advice and sentences throughout
      • Roger Lewin: transformed lecture transcripts into chapter drafts
      • Eric Chinski: editor who knew the book better than the author, making the final work an enjoyable collaboration
    • Essential Support Network
      • John Brockman: began as agent, became a trusted friend
      • Ran Hassin: provided advice and encouragement when most needed
      • Mary Himmelstein: offered valuable assistance throughout
    • Family's Crucial Role
      • Lenore Shoham: daughter who provided wisdom, sharp critical eye, and many "Speaking of" section sentences
      • Anne Treisman: wife whose steady support, wisdom, and endless patience prevented abandonment of the project
• Core Conclusion and Practical Takeaways
  • Core Ideas: The Two Systems and Their Flaws
    • System 1 runs the show: automatic, intuitive, and fast, but prone to systematic errors from heuristics and WYSIATI
    • System 2 is a lazy monitor: it often endorses System 1's flawed intuitions rather than exerting effort to correct them
    • Cognitive ease breeds gullibility: when things feel familiar or easy, System 1 accepts them as true without scrutiny
    • Losses loom larger than gains: the pain of losing is about twice as powerful as the pleasure of gaining, driving irrational risk choices
    • The remembering self tyrannizes decisions: we choose based on peak-end memories, not the total experience of our living self
  • Daily Practices: Taming Intuition
    • Recognize cognitive minefields: learn the diagnostic labels (anchoring, halo effect, availability cascade) to spot errors in yourself and others
    • Apply the outside view: when forecasting, start with base rates from a reference class, then adjust—don't build from your plan alone
    • Use the premortem: before finalizing a decision, imagine it has failed catastrophically and write the story of why
    • Slow down for high stakes: activate System 2 by frowning, questioning assumptions, and asking "What am I missing?"
    • Gather independent judgments first: before group discussion, collect opinions separately to avoid the first speaker's halo effect
  • Mindset Shifts: Redefining Rationality
    • Accept that you are not rational: rationality requires internal consistency and statistical thinking, which System 1 cannot deliver
    • Embrace algorithms over intuition: simple formulas and checklists consistently beat expert judgment in predictable environments
    • Distinguish skill from luck: genuine expertise requires a regular environment and immediate feedback; most long-term predictions are illusions
    • Frame decisions broadly: aggregate many risky choices into a policy ("you win a few, you lose a few") to overcome loss aversion
    • Design your choice architecture: use defaults, framing, and nudges to steer yourself and others toward better outcomes without coercion