Economic psychology

What this subject is

The study of how humans actually judge, decide and behave in economic contexts — as consumers, savers, taxpayers, investors, employees — against the normative benchmark of rational choice. Two research programs dominate its modern core: the heuristics-and-biases program (Tversky & Kahneman), which documents systematic deviations from probability theory and expected utility and culminates in prospect theory; and behavioural economics (Thaler), which carried those findings into economics via mental accounting, self-control and choice architecture. Both descend from Simon's bounded rationality, and both are challenged from inside psychology by Gigerenzer's ecological-rationality program and from metascience by the replication crisis.

What the package holds

Curated scaffolding your agent loads before it researches, so it starts from vetted ground rather than a cold search.

• Tier 1: 22
• Tier 2: 7
• Tier 3: 1
• Tier 4: 2

The questions and claims below are quoted from the package files.

Where the field disagrees

Each one carries real proponents on more than one side, so your agent cannot quietly pick a winner.

• Are heuristics defective shortcuts that produce systematic irrationality, or adaptive tools that are rational relative to their environments?

  4 named positions · partially reconciled in substance, unresolved in framing; both programs remain productive and every serious course teaches both

• Do nudges deliver policy-relevant effects, or is the celebrated literature an artefact of publication bias?

  4 named positions · converging on a sober middle — small, context-dependent, sometimes cost-effective; the strong early claims are retired

• After the replication crisis, which parts of the field's evidence base still stand?

  5 named positions · settled in outline — teach the core with confidence, the casualties as history, and the method lessons as content

• Does the famous parole study show extraneous factors dominating expert judgment?

  4 named positions · the strong reading is not tenable; retained as the package's standard exercise in interrogating a seductive result

• Do individual-level biases matter for market outcomes, or do aggregation, arbitrage and learning wash them out?

  4 named positions · unresolved and probably domain-specific; the honest answer is "depends on the institution"

Myths the package corrects

Widely held claims with the evidence that settles or bounds them.

• “Behavioural research proved that humans are irrational, full stop.”

  debunked-as-general-rule

  The research shows systematic, predictable deviations from specific normative models — a much narrower and more useful claim. The same deviations are patterned enough to be modelled (prospect theory is a theory, not a shrug), and Gigerenzer's program shows many heuristics are accurate in the environments they evolved for: recognition and take-the-best beat regression in real inference tasks, and presenting problems as natural frequencies dissolves much of the base-rate "irrationality". Whether a heuristic is rational is a question about the fit between mind and environment, not a verdict on the species. Kahneman himself describes System 1 as "generally very good" at what it does; the disagreement with Gigerenzer is over norms and framing, not over whether humans function.

• “Subtle primes — words about old age, images of money, a displayed flag — robustly steer real behaviour.”

  debunked-as-general-rule

  The flagship findings failed high-powered replication: the elderly-priming walking-speed effect disappeared with automated timing and blinded experimenters (doyen-2012); money and flag priming failed in the 36-lab Many Labs project (klein-many-labs-2014). Kahneman wrote an open letter in 2012 calling priming research a looming "train wreck" and later stated publicly that he had placed too much faith in underpowered studies and that readers should discount that chapter of his book. Within-person semantic priming in cognitive psychology (faster word recognition after related words) remains solid — the casualty is the claim that incidental primes appreciably steer downstream social and economic behaviour.

• “Nudges reliably produce large behaviour change at negligible cost — a free lunch for policy.”

  contested-strong-reading-unsupported

  The two best 2022 estimates agree the promise was oversold and disagree only on how much survives. Maier et al. (maier-2022) show the published nudge literature is so publication-biased that the bias-corrected average effect is indistinguishable from zero. DellaVigna and Linos (dellavigna-linos-2022), using every trial from two US government nudge units — published or not — find real average effects of about 1.4 percentage points, roughly one-sixth of what the academic literature advertises, still often cost-effective because nudges are cheap. Defaults in savings remain the genuine success story; "nudges reliably produce large effects" is not supported by either estimate.

• “Mainstream economics assumes Homo oeconomicus as a literal description of every human being, so any observed bias refutes economics.”

  debunked

  Rational choice in economics is mostly an as-if modelling device (Friedman's 1953 methodology essay) and a benchmark, not an anthropological claim — and economists themselves supplied the deviations: Simon (bounded rationality), Katona (sentiment), Thaler (anomalies published in economics journals), with Nobels awarded for exactly this work in 1978, 2002 and 2017. Conversely, individual bias does not automatically translate into aggregate irrationality: Vernon Smith's double-auction experiments and Gode & Sunder's zero-intelligence traders show market institutions can produce near-rational outcomes from non-rational participants, and List's field experiments show experience attenuating the endowment effect. The serious question is not "is Homo oeconomicus true" but "which institutions make which deviations matter".

• “The hungry-judges study proved that judicial decisions are driven by whether the judge has eaten.”

  debunked-as-general-rule

  The pattern is real in the data; the fatigue interpretation is not established. Weinshall-Margel and Shapard showed case ordering was non-random — unrepresented prisoners tend to appear later in sessions — which reproduces the drop without any hunger. Glöckner added the magnitude argument: an effect that moves approval from 65 percent to near zero would be vastly larger than any known cognitive effect, which should itself have triggered doubt. The study now functions as the field's standard exercise in interrogating seductive findings: check the mechanism, check the ordering, check whether the effect size is even plausible.

Related topics

• Game theory
• Austrian economics
• Habit formation