Thinking, Fast and Slow

The two characters of the story

System 1 continuously produces impressions, feelings, intentions, and intuitions. System 2 can endorse them, modify them, reject them, or turn them into deliberate choices.

When you see an angry face, understand a simple sentence, or recoil from a sudden sound, System 1 acts immediately. When you calculate 17 × 24, compare insurance policies, or check whether an argument is logically valid, System 2 must concentrate.

System 2 believes itself to be the main character, but much of what it thinks and chooses begins as a suggestion from System 1.

Attention is a limited budget

Effortful thinking draws from a finite pool of attention. Complex mental operations interfere with one another because they compete for the same resource.

You may be able to walk and talk at once, but a difficult calculation can cause you to stop walking. You can drive on an empty road while conversing, but navigating dangerous traffic demands concentration.

This produces an important principle:

You cannot assume that a person who possesses relevant knowledge will use it while distracted, hurried, emotionally aroused, or cognitively overloaded.

Mental performance depends not only on intelligence but on whether adequate attention is available at the moment of judgment.

System 2 is a lazy controller

System 2 usually invests only as much effort as appears necessary. This conserves energy, but it also allows intuitive mistakes to pass unchallenged.

Kahneman's famous bat-and-ball problem demonstrates this:

A bat and ball cost $1.10 together. The bat costs $1 more than the ball. How much does the ball cost?

System 1 quickly suggests ten cents. But if the ball costs ten cents, the bat costs $1.10, making the total $1.20. The correct answer is five cents.

The wrong answer feels right because it arrives fluently. Detecting the mistake requires System 2 to interrupt intuition and verify the arithmetic.

Intelligence does not guarantee vigilance. Intelligent people can use their abilities to justify an intuitive answer instead of examining it.

Self-control and cognitive control are connected

System 2 is responsible for monitoring behavior as well as solving problems. It helps suppress inappropriate responses, sustain attention, follow rules, and resist temptation.

When attention is absorbed elsewhere, self-monitoring may weaken. Fatigue, stress, time pressure, intoxication, and cognitive overload make people more likely to accept the first plausible answer or act upon an immediate impulse.

The broader lesson is that judgment is partly situational. A person may make different decisions not because their values changed but because their capacity for deliberate control changed.

The associative machine

System 1 operates through networks of association. One idea activates related ideas, emotions, memories, bodily reactions, and impulses.

A word such as "vomit," for example, does more than communicate information. It may produce facial tension, disgust, memories, and a desire to move away. These responses arise together as an associative pattern.

System 1 continually constructs a coherent interpretation of the present by linking what it sees with what memory makes available.

This process is powerful but permissive. Associations do not need to be logically proven before influencing judgment.

Priming shapes thought and behavior

Recent exposure to an idea can make related ideas easier to access. That accessibility may influence later interpretations or actions without the person recognizing the source.

The central principle is that cognition is embodied and associative. Thoughts can influence physical responses, and bodily states can influence thought.

The conscious mind therefore does not operate in isolation. Its judgments are partly prepared by context, language, mood, memory, and cues it may not consciously register.

Cognitive ease makes ideas feel true

The mind monitors how easily it is processing information.

Cognitive ease can be created by:

• Repetition
• Familiarity
• Clear presentation
• Good mood
• Prior exposure
• Simple language
• A coherent story

When processing feels easy, people are more likely to experience familiarity, confidence, liking, and truth. When processing is strained, they become more vigilant, analytical, and suspicious.

This explains the illusion of truth: repeated statements feel more believable even when repetition provides no new evidence.

A polished presentation may therefore influence how credible an argument feels. Fluency is easily mistaken for validity.

System 1 maintains a model of normality

System 1 continuously learns what normally follows what. It detects surprise by comparing events with its implicit expectations.

Once an unusual event occurs several times, it can quickly become the new normal. The mind adjusts its expectations without requiring deliberate analysis.

Surprise also triggers causal interpretation. When something unexpected happens, System 1 searches for a reason — even when the event may have resulted from chance.

Humans are natural storytellers because causal explanations are more psychologically satisfying than unexplained randomness.

The mind jumps to conclusions

System 1 can form confident judgments from limited evidence. It emphasizes the coherence of the available story rather than the quantity or quality of the evidence supporting it.

This is WYSIATI:

What you see is treated as though it were all that exists.

WYSIATI helps explain:

• Overconfidence
• The halo effect
• Premature conclusions
• Neglect of missing evidence
• Insensitivity to sample size
• Excessive faith in personal experience

The mind can create a perfect story from dangerously incomplete information because missing information does not automatically announce itself.

The halo effect creates artificial consistency

A strong positive or negative impression in one area influences unrelated judgments.

If someone appears confident and attractive, an observer may also assume that the person is intelligent, trustworthy, and competent. If the first information received about someone is positive, later ambiguous information is interpreted more favorably.

Order therefore matters. Early impressions create a narrative into which later evidence is fitted.

One remedy is to gather independent judgments before group discussion. If everyone hears the leader's opinion first, subsequent assessments become correlated and the group loses valuable independent information.

System 1 performs basic assessments

The mind automatically evaluates many qualities:

• Threat
• Dominance
• Similarity
• Likeability
• Emotional intensity
• Causal coherence
• Whether something is normal or surprising

These assessments are often useful, but they can influence judgments that should depend on different information.

Kahneman calls this overflow the mental shotgun: when the mind is asked one question, it automatically computes answers to related questions as well. Those unintended assessments can contaminate the answer.

Intensity matching crosses dimensions

System 1 can match intensity across unrelated scales.

A severe crime feels as though it deserves a severe punishment. A highly impressive student feels destined for a highly impressive career. An emotionally powerful threat feels highly probable.

This operation is intuitive, not statistical. It substitutes correspondence in emotional intensity for a defensible quantitative judgment.

Difficult questions are replaced with easier ones

When System 1 cannot answer a difficult target question quickly, it may answer an easier heuristic question without announcing the substitution.

System 2 often accepts the substituted answer because it arrives quickly and coherently. A powerful corrective question is:

What question am I actually answering?

The affect heuristic

People often allow their general emotional reaction to determine their estimates of benefits and risks.

If they like something, they tend to see:

• High benefits
• Low costs
• Low risks

If they dislike it, they tend to see:

• Few benefits
• High costs
• High risks

Emotion creates a package of mutually supporting judgments. New information may then be interpreted to preserve the original feeling.

The law of small numbers

People expect small samples to resemble the population from which they were drawn more closely than they actually do.

A few successful trades may look like evidence of skill. Several failures under a new manager may look like proof of incompetence. Three strong product reviews may seem decisive.

But small samples produce extreme results more frequently because they contain more random variation.

The mind believes in a nonexistent "law of small numbers": it expects even limited evidence to display the stability of a large sample.

Randomness does not look random

True randomness naturally produces clusters, streaks, repetitions, and apparent patterns. Yet people often expect random sequences to alternate more regularly than they really do.

This gives rise to the gambler's fallacy: after several occurrences of one outcome, the opposite outcome feels "due," even when the events are independent.

The same misunderstanding works in reverse when people interpret a streak as evidence of momentum, special talent, or a "hot hand" without sufficient data.

Anchoring pulls estimates toward an initial value

An anchor is a number encountered before an estimate is made. Judgments remain biased toward it even when it is arbitrary or obviously imperfect.

Anchoring occurs through two processes:

1. Insufficient adjustment: System 2 starts from a value and fails to move far enough away.
2. Associative activation: System 1 searches for information compatible with the anchor.

Anchors affect negotiations, valuations, damages, prices, forecasts, and estimates.

The first number introduced into a discussion can reshape the entire field of thought. Defending against an anchor requires deliberately considering reasons the anchor may be wrong and constructing an independent estimate.

Availability substitutes recall for frequency

The availability heuristic estimates the frequency or probability of an event by how easily examples come to mind.

Events become available when they are:

• Recent
• Vivid
• Dramatic
• Emotional
• Personally experienced
• Repeated by the media
• Easy to imagine

Availability is sometimes informative: common events often are easier to recall. But memory is also shaped by attention and publicity. A vivid risk can dominate judgment while a larger but less dramatic danger is ignored.

Availability also changes self-assessment

People infer qualities from how easily they can retrieve examples.

If asked to recall a few occasions when they behaved assertively, examples come easily and they may judge themselves assertive. If asked to recall many occasions, the search becomes difficult and they may conclude they are less assertive — even though they produced more evidence.

The experience of retrieval can outweigh the content retrieved.

The mind asks not only, "What examples did I find?" but also, "How hard was it to find them?"

Availability cascades magnify public fear

A vivid event receives attention. Attention produces public concern. Concern generates more media coverage. Increased coverage makes the event even easier to recall, which further increases concern.

The resulting collective concern may become disconnected from the underlying probability.

Public risk perception is therefore shaped not only by facts but by mutually reinforcing cycles of attention, emotion, and repetition.

Representativeness replaces probability with resemblance

When judging whether someone or something belongs to a category, people often compare it with a stereotype.

The classic example describes Steve as quiet, orderly, and detail-oriented, then asks whether he is more likely to be a librarian or farmer. The description resembles the stereotype of a librarian, but there are far more farmers. The base rate matters.

Representativeness answers:

How much does this case resemble my mental picture?

Probability asks:

Given the base rate and the strength of the evidence, how likely is the case?

These questions are not the same.

Base rates should govern weak evidence

When specific evidence is unreliable, predictions should remain close to the statistical base rate. As diagnostic evidence becomes stronger, the estimate can move further away from that baseline.

A disciplined forecast therefore combines:

• The frequency of the outcome in the relevant population
• The reliability and relevance of the case-specific evidence

System 1 naturally emphasizes vivid individual evidence and neglects abstract statistics. Good judgment restores the statistical background.

The conjunction fallacy: plausible stories outrank logic

The Linda problem describes a woman whose personality and history resemble a social activist, then asks whether she is more likely to be:

1. A bank teller, or
2. A bank teller active in the feminist movement

The second description feels more representative, but it cannot be more probable. Every feminist bank teller is also a bank teller; adding a condition can only reduce or preserve probability.

This is the conjunction fallacy. The richer description tells a better story, and System 1 confuses plausibility with probability.

More detail can make a scenario more convincing while making it less likely.

Causes have more psychological power than statistics

The mind prefers causal stories about individuals to abstract statistical information.

A statistical base rate may be ignored when a vivid description is available. Yet the same information can influence judgment if it is translated into a causal stereotype or concrete example.

This creates an educational difficulty: learning a general fact does not necessarily change behavior. People often learn more deeply when they see how an individual case contradicts their intuitive assumptions.

Regression toward the mean

Whenever performance combines skill and luck, extreme outcomes are likely to be followed by less extreme outcomes.

An athlete who performs extraordinarily well probably benefited from both skill and unusually favorable circumstances. On the next attempt, the favorable circumstances are unlikely to be equally strong, so performance tends to move closer to the average.

Regression is often mistaken for causation:

• Praise appears to make performance worse because unusually good performance naturally declines.
• Criticism appears to improve performance because unusually poor performance naturally improves.
• A treatment appears effective because people often seek treatment when symptoms are unusually severe and symptoms later return toward normal.

Nothing needs to cause this movement. It is a statistical consequence of imperfect correlation.

Extreme predictions should be moderated

System 1 makes predictions by matching the extremity of the evidence to the extremity of the forecast. An outstanding interview inspires an outstanding career prediction.

Kahneman's more disciplined method is:

1. Estimate the relevant baseline.
2. Form an intuitive prediction.
3. Estimate how reliable the evidence is.
4. Move the intuitive prediction toward the baseline according to that reliability.

Weak evidence deserves substantial regression toward the mean. Strong evidence permits a more extreme forecast. This often produces predictions that feel insufficiently bold but are more defensible.

The illusion of understanding

After an event occurs, the mind constructs a causal story explaining why it happened. Because the story is coherent, the outcome begins to feel inevitable.

This is the narrative fallacy: complex events involving luck, hidden variables, and unpredictable interactions are compressed into tidy explanations.

Success stories are especially deceptive because they focus on visible winners. They rarely include equally promising people or companies that failed because of factors absent from the narrative.

A coherent past creates the illusion that the world is more understandable than it is.

Hindsight changes the memory of former beliefs

After learning an outcome, people struggle to reconstruct what uncertainty genuinely felt like beforehand. They remember the event as more predictable and their original judgment as closer to the truth.

Hindsight supports two related errors:

• Outcome bias: judging a decision by how it turned out.
• Hindsight bias: believing the outcome should have been foreseen.

A sound decision can produce a bad outcome, and a reckless decision can succeed through luck. Decision quality must be evaluated using the information available when the decision was made.

The illusion of validity

Confidence is often generated by the coherence of a story, not by its predictive accuracy.

If available cues fit together neatly, the mind experiences confidence. But highly coherent evidence can be redundant: several indicators may all reflect the same underlying factor and therefore provide less independent information than they appear to.

Subjective confidence answers:

How coherent is the story in my mind?

It does not necessarily answer:

How predictable is the world represented by that story?

The strength of conviction should not be confused with the quality of evidence.

Confidence is not proof of expertise

Professionals can feel highly confident in domains where outcomes remain largely unpredictable. Experience supplies stories and familiar patterns, but experience alone cannot create reliable intuition.

Kahneman distinguishes the confidence of the judgment from the validity of the environment. To assess an expert, ask:

• Does the environment contain stable patterns?
• Are those patterns discoverable?
• Has the person had extensive practice?
• Was feedback rapid, clear, and accurate?
• Did the person have genuine opportunities to learn?

Without these conditions, confidence may reflect coherence rather than skill.

When expert intuition can be trusted

Reliable intuition develops when two conditions are present:

• A sufficiently regular environment
• Prolonged practice with valid feedback

Chess, certain medical diagnoses, firefighting, and many practiced physical skills can meet these conditions. The expert sees a cue, memory recognizes the pattern, and an appropriate response appears automatically.

Intuition is therefore not magic. It is recognition.

By contrast, long-term political forecasts, many market predictions, and complex business outcomes occur in environments that may be too irregular for stable intuitive expertise.

Simple formulas often outperform unaided judgment

Mechanical rules frequently outperform human judges in prediction — even when the judges possess more information.

Humans are inconsistent. The same person can interpret identical evidence differently depending on mood, fatigue, context, or recent experience. Formulas apply the same rule every time.

A useful model need not be mathematically sophisticated. Often the improvement comes from:

• Selecting a few relevant variables
• Scoring them separately
• Assigning consistent weights
• Combining them mechanically
• Preventing a vivid impression from dominating

Kahneman does not argue that judgment is useless. Human judgment is valuable when selecting inputs, recognizing exceptional circumstances, and designing the rule. But final integration often benefits from structure.

Structure improves interviews and hiring

Unstructured interviews invite halo effects, substitution, and excessive confidence. Interviewers form a global impression and then reinterpret answers to fit it.

Kahneman recommends:

1. Identify a limited set of traits relevant to performance.
2. Create factual questions for each trait.
3. Score each dimension independently.
4. Avoid forming an overall judgment too early.
5. Combine the scores systematically.
6. Add holistic judgment only after the structured assessment.

This preserves information that would otherwise be swallowed by the interviewer's first impression.

The inside view causes the planning fallacy

The inside view focuses on the details of the current project:

• The plan
• The team
• The expected steps
• The imagined best path
• The obstacles currently visible

This produces optimistic estimates because the plan is usually a best-case scenario. It cannot represent every possible delay, failure, illness, conflict, or unexpected complication.

The outside view asks what happened in comparable cases. Reference-class forecasting begins with actual completion times, costs, and failure rates from similar projects, then adjusts cautiously for genuine differences.

The outside view is usually uncomfortable

Project teams resist the outside view because their project feels unique. They possess detailed inside knowledge and believe the details justify optimism.

But uniqueness itself is often part of the illusion. Every failed project had an inside story explaining why it would succeed.

The outside view deliberately ignores much of the project's narrative at first. It begins with the base rate and forces the plan to earn every departure from it.

Optimism is both useful and dangerous

Optimism helps people act, persevere, mobilize others, and tolerate uncertainty. It contributes to entrepreneurship and innovation.

But optimistic people also tend to:

• Overestimate their abilities
• Underestimate competitors
• Underestimate costs and delays
• Exaggerate personal control
• Neglect base rates
• Focus on goals while ignoring obstacles
• Interpret commitment as evidence of likely success

Optimism may benefit society by encouraging experimentation even when many individual ventures fail. The same bias can be socially productive and personally costly.

The premortem exposes hidden threats

Before final approval of an important plan, Kahneman recommends an exercise developed by Gary Klein:

Imagine that the decision has been implemented and, one year later, it has failed disastrously. Write the history of that failure.

The premortem legitimizes doubt. Team members who might otherwise appear disloyal are invited to identify threats. It can reveal:

• Competitor responses
• Coordination failures
• Unrealistic timelines
• Hidden dependencies
• Weak assumptions
• Regulatory obstacles
• Customer resistance
• Problems known privately but not voiced publicly

It does not eliminate uncertainty, but it reduces the suppression of relevant information by group enthusiasm.

Traditional utility theory misses the reference point

Classical theories often assume that people evaluate their final state of wealth. Kahneman and Tversky argue that people experience gains and losses relative to a reference point.

Someone receiving a $10,000 bonus may feel delighted if expecting nothing and disappointed if expecting $20,000. The final wealth can be identical while the emotional outcome differs. The evaluation depends upon:

• What the person previously had
• What they expected
• What they believe they are entitled to
• What others received
• Which outcome became the reference point

Human value is reference-dependent.

Prospect theory has three central features

Reference dependence. Outcomes are perceived as gains or losses relative to a reference point, not simply as final states.

Diminishing sensitivity. The subjective difference between $100 and $200 feels greater than the difference between $1,100 and $1,200. Sensitivity decreases as one moves farther from the reference point.

Loss aversion. A loss generally has more psychological impact than an equivalent gain.

These features produce the prospect theory value curve:

• Concave for gains
• Convex for losses
• Steeper on the loss side
• Defined around a reference point

The emotional geometry of gains and losses shapes risk-taking.

Losses loom larger than gains

The possibility of losing something creates stronger motivation than the possibility of gaining an equivalent amount. This influences:

• Investment decisions
• Negotiations
• Employment contracts
• Pricing
• Political reforms
• Territorial disputes
• Consumer behavior
• Personal attachments

Loss aversion favors the status quo. Once something is mentally coded as "mine," surrendering it feels like a loss that must be compensated by a larger gain.

The endowment effect

People often demand more to give up an object they own than they would have paid to acquire it.

Ownership changes the reference point. Before ownership, the object is a possible gain. After ownership, parting with it becomes a loss.

The endowment effect is strongest for items held for use or attachment. It may be weaker for professional traders because goods intended for exchange are not encoded in the same way.

The deeper principle is that preferences are partly constructed by current possession. They are not always stable values waiting to be discovered.

Loss aversion preserves the status quo

Those who stand to lose from a change usually fight harder than those who stand to gain. This creates an asymmetry:

• Benefits are dispersed and weakly defended.
• Losses are concentrated and fiercely resisted.
• Concessions appear more painful to the giver than valuable to the receiver.
• Existing entitlements become psychologically protected.

Reforms can therefore fail even when total gains exceed total losses. Aggregate value does not erase the emotional force of individual losses.

Fairness is defined around reference transactions

People judge economic behavior partly according to customary terms.

A business may be considered entitled to preserve its normal profit when costs rise. But using a temporary shortage solely to impose a large price increase may be experienced as an unfair exploitation of customers.

Fairness becomes an economic constraint because customers and employees may punish violations, even at personal cost. Reference points therefore influence not only private emotion but markets, contracts, and social norms.

The fourfold pattern of risk

Risk preferences change according to whether outcomes involve gains or losses and whether probabilities are high or low.

This pattern explains why the same person can buy both insurance and lottery tickets without experiencing inconsistency.

The possibility and certainty effects

A change from zero probability to a small probability has disproportionate psychological impact. A previously impossible outcome becomes possible.

Likewise, a change from high probability to certainty is valued more than an equivalent increase that does not eliminate uncertainty. These effects help explain why people:

• Pay for peace of mind
• Overvalue guarantees
• Overpay to eliminate tiny risks
• Become excited by tiny chances of enormous gains
• Reject favorable settlements when a small chance of avoiding all loss remains

Probability is psychologically weighted rather than treated linearly.

Rare events are often overweighted in descriptions

When rare outcomes are explicitly described, vividness and focused attention can make them loom too large. People may overweight:

• A tiny chance of catastrophe
• A tiny chance of a jackpot
• A frightening scenario
• A vivid medical outcome
• A memorable investment success

The effect becomes stronger when the event is emotionally charged or easily imagined.

But rare events can also be neglected when learned only through experience. If the event does not occur in a limited personal sample, it may effectively disappear from judgment. Thus rare events may be either overweighted or ignored depending on how information is encountered.

Denominator neglect distorts risk

A statement such as "1,286 people out of every 10,000" often creates more fear than "12.86%," even though the information is equivalent.

The first format presents visible victims. System 1 responds to the image of individuals rather than calculating the proportion.

Risk communication is therefore sensitive to format. Frequencies, percentages, survival rates, and mortality rates may create different emotional responses while describing the same reality.

Broad framing improves repeated decisions

People naturally evaluate decisions one at a time. This narrow framing magnifies loss aversion.

An investor who watches every movement experiences many painful losses. An organization whose managers evaluate each uncertain project separately may reject a portfolio of favorable risks. Broad framing considers the aggregate:

• Some decisions will lose.
• Others will win.
• What matters is whether the policy is favorable across many repetitions.

The correct unit of evaluation is often not the individual bet but the portfolio of similar decisions.

Risk policies reduce emotional inconsistency

A risk policy decides in advance how a class of recurring choices will be handled. Examples include:

• Do not buy extended warranties.
• Accept small favorable risks when financially tolerable.
• Rebalance investments on a fixed schedule.
• Judge a strategy after a meaningful sample, not after each outcome.
• Do not alter a sound policy because of one vivid result.

Policies protect decisions from the mood created by the most recent gain or loss.

Mental accounts divide money into artificial categories

People maintain separate psychological accounts for different goals, purchases, investments, and sources of money. This can produce inconsistent choices:

• Treating a tax refund differently from earned income
• Refusing to sell a losing investment while readily buying a better one elsewhere
• Continuing a failing project because of past spending
• Taking greater risks with unexpected gains
• Preserving one account while borrowing at high cost in another

Money is economically interchangeable, but it is not psychologically interchangeable.

The sunk-cost fallacy

Once time, money, or identity has been invested, abandoning a failing course feels like accepting a loss.

But past costs cannot be recovered. The rational question is:

Given my options today, which course has the best future prospects?

The sunk-cost fallacy substitutes another question:

How can I avoid admitting that the earlier investment was wasted?

Organizations are especially vulnerable when the person responsible for the original decision also decides whether to continue it.

The disposition effect

Investors often sell profitable holdings too early and retain losing holdings too long.

Selling a winner allows the investor to record a success. Selling a loser forces the closing of a mental account with a painful result.

The purchase price becomes a reference point even when it has little bearing on the asset's future value. A cleaner question is:

If I did not already own this investment, would I buy it today at its present price?

Regret depends upon action, responsibility, and alternatives

People anticipate regret when deciding. The possibility of later blaming oneself can encourage inaction, conformity, and excessive caution. Regret tends to intensify when:

• The person departed from a normal policy
• The action was unusual
• A clear alternative was available
• The outcome is easy to compare with what might have happened
• The person feels personally responsible

Consistent policies reduce regret because a poor outcome can be understood as one instance within a defensible process.

Preferences reverse across contexts

People may assign different values to the same objects depending on whether they evaluate them separately or jointly.

In separate evaluation, easily interpretable attributes dominate. In joint evaluation, direct comparison makes previously neglected information visible.

This means preferences are often constructed in the act of evaluation. They are not always fixed, internally consistent, or independent of context.

Framing changes choices without changing outcomes

Equivalent descriptions can produce different decisions.

A medical procedure described by its survival rate may be received more favorably than the same procedure described by its mortality rate. A choice framed as preventing losses may motivate more strongly than one framed as creating equivalent gains. Frames determine:

• The reference point
• What counts as a gain
• What counts as a loss
• Which features receive attention
• Which emotions are activated

System 2 may know that two frames are logically equivalent while System 1 continues to react differently.

Defaults are powerful frames

A default feels normal, recommended, and effortless. Departing from it requires action and creates greater responsibility. Defaults strongly influence:

• Retirement savings
• Insurance selection
• Subscriptions
• Organ donation
• Privacy settings
• Employee benefits

Choice architecture matters because actual humans do not approach every choice with unlimited attention and perfect neutrality.

The experiencing self and remembering self

Kahneman distinguishes between:

• The experiencing self, which lives through each moment
• The remembering self, which evaluates the episode afterward and makes future decisions

These selves do not measure experiences in the same way. The experiencing self is sensitive to the quality of every moment. The remembering self compresses the experience into a story, emphasizing selected moments and the ending.

This creates a conflict: the self that chooses may not choose what produces the greatest total lived well-being.

The peak-end rule

Memory's evaluation of an episode is influenced heavily by:

• The most intense moment
• The way the experience ended

The total duration receives surprisingly little weight.

A painful medical procedure may be remembered more favorably if it ends with a period of reduced pain — even if that extension adds more total discomfort.

The remembering self prefers a better story, while the experiencing self would prefer less total suffering.

Duration neglect

When people evaluate a past experience, its length may matter far less than expected.

A longer unpleasant experience can receive a better retrospective evaluation if its ending is gentler. A longer enjoyable experience may add little to memory if it does not change the peak or ending.

This reveals that remembered utility and experienced utility are different measures.

Life is remembered as a story

The remembering self organizes life into episodes with beginnings, important moments, turning points, and endings. It cares about:

• Memorable peaks
• Significant changes
• Goals achieved or lost
• How episodes conclude
• Whether events form a coherent narrative

Duration has less narrative importance. A long life is not automatically remembered as better than a somewhat shorter life if the added years damage the story's ending.

This "tyranny of the remembering self" influences decisions because memory chooses which experiences to repeat.

Vacations reveal the conflict between the selves

Suppose you knew that every photograph would be destroyed and your memory of a vacation erased afterward. Would you choose the same trip?

The question separates:

• The value of living the experience
• The value of remembering and narrating it

Some choices are made primarily for anticipated memories rather than moment-to-moment experience. Photographs, stories, milestones, and memorable peaks are designed for the remembering self.

Neither self is declared illegitimate. The difficulty is that their interests can diverge.

Experienced well-being should be measured in time

Global questions such as "How happy are you with your life?" produce evaluations from the remembering self.

Experienced well-being asks how people feel during the actual moments of their days. Kahneman discusses methods that reconstruct daily episodes and measure the emotions attached to them. The two forms of well-being differ:

• Life satisfaction is an evaluation of one's life.
• Experienced happiness is the balance and quality of moment-to-moment experience.

A person may judge life positively because important goals were achieved while spending many hours in stress. Another may experience pleasant days while judging life as insufficiently successful.

Attention determines experienced reality

Emotional circumstances affect experience most when they occupy attention.

Many life conditions become less emotionally powerful as people adapt and stop attending to them. The condition remains objectively present, but it no longer dominates consciousness.

Kahneman's practical principle is:

Nothing in life is as important as you think it is while you are thinking about it.

This is the focusing illusion. When attention is concentrated on one feature — income, weather, status, health, appearance — it appears to determine life satisfaction more than it usually does across daily life.

Adaptation is selective

People adapt to many circumstances because attention moves elsewhere. But adaptation is incomplete when a condition continually demands attention.

Chronic pain, persistent noise, severe depression, exhausting commutes, and time pressure may repeatedly return themselves to consciousness. They can therefore exert sustained effects on experienced well-being.

The effect of a circumstance depends partly on whether it remains an active focus in daily experience.

Goals shape life satisfaction

The remembering self evaluates life partly by comparing achievements with aspirations.

Goals give direction, but they also create reference points. A person can achieve much and remain dissatisfied if the reference standard rises equally quickly. Another may judge life positively because outcomes exceeded modest expectations. Life satisfaction therefore depends not only on circumstances but on:

• Aspirations
• Comparisons
• Remembered achievements
• Cultural standards
• Current mood
• Which part of life receives attention during evaluation