A thirteen-slide synopsis of the heuristics-and-biases program, prospect theory, and the descendants of Simon, Kahneman, Tversky, and Thaler.
For most of the twentieth century, mainstream economics modeled decision-makers as homo economicus: agents with stable preferences, unlimited computational power, and the discipline to maximize expected utility under uncertainty1.
The framework is elegant. It is also empirically wrong. Real humans systematically violate its axioms — not at the margins, but at the center.
The deviations from rational choice are not random noise. They are structured, predictable, and exploitable.
Behavioral economics studies that structure — what Kahneman called the "map of bounded rationality."
Simon's central insight: cognition is a scarce resource. An agent searching for the optimal apartment, spouse, or chess move would never finish. Real decision-making terminates at aspiration thresholds — "good enough" beats "best."3
Simon won the 1978 Nobel for showing that institutions and routines are not failures of rationality — they are how rationality survives finite minds in infinite worlds.
When a satisficer finds options easily, the threshold rises. When search drags, it falls. This adaptive thermostat resembles real consumer behavior far better than maximization does.
The 1974 Science paper catalogued three master heuristics that humans deploy when judging probabilities. Each is cheap, fast, and reliable in the ancestral environment — and each produces characteristic errors when transplanted to modern statistical questions5.
Estimate frequency by ease of recall. Plane crashes feel common; bathtub deaths do not. Cause: vivid retrieval, not base rates.
Judge category membership by similarity. Linda the bank teller. The conjunction fallacy. Base-rate neglect.
Adjust insufficiently from a numerical starting point — even arbitrary ones. Spinning a wheel of fortune before estimating UN African membership shifts answers in the direction of the wheel.
The 1979 Econometrica paper has been cited over 70,000 times. Its three claims:7
v(x) = xα for x ≥ 0 ; v(x) = −λ(−x)β for x < 0. With α,β ≈ 0.88, λ ≈ 2.25.
Richard Thaler observed that households partition money into psychological "accounts" — rent, groceries, vacation, the kids' college fund — and behave very differently across them9. The same $1,000 is treated as untouchable in one account and trivially spendable in another.
You arrive at the theater and discover you've lost your $50 ticket. Most people will not buy a replacement. But if you'd lost a $50 bill on the way, most people will still buy the ticket. Same wealth loss; different mental account.
Mental accounting explains why people simultaneously hold credit-card debt at 22% and a savings account at 4%. The accounts don't talk.
| Account | Source | Treatment |
|---|---|---|
| Salary | Earned | Saved, careful |
| Bonus | Windfall | Spent, indulgent |
| Tax refund | "Found" | Spent, often impulse |
| Inheritance | Solemn | Held, untouched |
| Casino chips | "House money" | Risk-loving |
The "house money effect" — people gamble more aggressively with winnings than with their original stake — is a direct prediction.
The classic mug experiment (Kahneman, Knetsch & Thaler, 1990): half of a class is randomly given a coffee mug. Buyers are asked the maximum they'd pay for one; sellers the minimum they'd accept to give theirs up.11
Standard theory predicts these prices should match (Coase). They do not. Sellers ask about twice what buyers offer — instantly, on objects owned for under five minutes.
Giving up the mug is coded as a loss. Acquiring one is coded as a gain. Loss aversion (λ ≈ 2) does the rest.
Status-quo bias plus the cognitive cost of switching produces enormous gaps between opt-in and opt-out regimes — even when the friction is one click or one form.13
Germany (opt-in): ~12% consent rate. Austria (opt-out): ~99% consent rate. The countries are otherwise comparable. The default did the work.
Madrian & Shea (2001): switching one US firm from opt-in to automatic enrollment raised participation from 49% to 86%. Crucially, employees did not subsequently opt out — they stayed.
The neoclassical model discounts the future exponentially: the trade-off between today and tomorrow has the same shape as between day 100 and day 101.
Empirically, humans discount hyperbolically: the near future is weighted hugely more than the medium future, but the medium and far futures are weighted similarly.15
Asked today: "$110 in 31 days vs. $100 in 30 days?" → most pick $110. "$110 tomorrow vs. $100 today?" → many flip and pick $100.
The same agent prefers patience-in-the-future and impatience-now. Hence New Year's resolutions, gym memberships, retirement under-saving, and procrastination.
A nudge is any change in the choice architecture that alters behavior in a predictable way without forbidding any option or significantly changing economic incentives.17 Cafeterias placing salad at eye level is a nudge; banning fries is not.
People remain free to choose. But because some default must exist, a designer should pick the one that, by the chooser's own lights, makes them better off.
Founded 2010. A single sentence added to UK tax-arrears letters — citing local compliance rates — yielded an estimated £200M+ in additional revenue within two years.
"Sludge" — the same architectural levers used against consumers (dark patterns, hard-to-cancel subscriptions). Thaler himself has warned about the asymmetry.
Behavioral economics borrowed heavily from social psychology, and inherited some of its problems when the replication crisis arrived.19
Heinrich, Heine & Norenzayan (2010) showed many "universal" findings rest on WEIRD samples (Western, Educated, Industrialized, Rich, Democratic). Loss aversion magnitudes vary across cultures; ultimatum-game offers vary by orders of magnitude across small-scale societies.
Loss aversion, the endowment effect, default effects, hyperbolic discounting, and base-rate neglect have all replicated robustly in pre-registered, well-powered tests. The core program is intact; the periphery is contested.
Over 200 government "nudge units" worldwide. Behavioral testing of letters, forms, and defaults is routine in tax authorities, public-health agencies, and pension regulators.
The US CFPB (2011) was designed with explicit behavioral-economics premises: disclosure formats are tested, late-fee architectures are reformed, mortgage docs simplified. Auto-enrollment is now a default across OECD pension systems.
Acorns (round-up saving), Digit and Chime (auto-saving by present-bias workaround), Robinhood (gamified trading — negative nudge), Apple Health closing rings. Behavioral architecture is now product design.
Default green-energy tariffs in Germany; smoking-cessation contracts (commitment devices); social-comparison utility bills shown to reduce consumption ~2%.
"We are not standard-issue rational agents — and at last our institutions are starting to admit it."
» Daniel Kahneman — Thinking, Fast and Slow
» Richard Thaler — Nudge & behavioral finance
See also: Dan Ariely TED talks, Cass Sunstein lectures on libertarian paternalism, and the LSE / NYU behavioral economics lecture series.
The achievement of behavioral economics is not that it proved economists wrong. It is that it gave the discipline a more accurate subject.