Revealed Preference

Version

v1 · 2026-08-24 · History

Prime #

1147

Origin domain

Economics & Finance

Subdomain

choice theory → Economics & Finance

Core Idea

Revealed preference is the structural commitment to infer latent valuations from observed choices rather than from solicited reports. The defining move is to treat the choice trace — the record of which alternative an agent picked when others were available — as the load-bearing evidence about what the agent values, and to treat any verbal report ("I prefer X to Y") as a separate and possibly inconsistent channel that the analyst is not obligated to trust.

The commitment has three tight components. The first is an epistemic asymmetry: the choosing act is harder to fake than the reporting act, because choosing carries an opportunity cost while reporting does not. The second is a substitution claim: under reasonable behavioral axioms — consistency, weak transitivity, no spite or self-deception — the chosen alternative carries at least as much information as a sincere report and strictly more than a distorted one. The third is an inversion procedure: from a sufficient record of choices across varying menus, the analyst recovers a preference structure (or utility function, or value ranking) that rationalizes the choices. That procedure has a known failure mode — intransitive cycles imply that no rationalizing preference exists — and the failure is itself diagnostic.

The pattern is recognizable wherever an agent's acted behavior diverges from their stated behavior and the analyst must decide which to credit. Naming the structure as revealed preference commits the analyst to credit the act over the statement, but with eyes open to the framing conditions under which that commitment is sound — and to the conditions under which the inversion breaks and the stated channel may in fact be the better evidence.

How would you explain it like I'm…

Watch, Don't Ask

If a kid says "I like apples best" but always grabs the cookie when both are out, watch what they grab, not what they say. What you pick when you could pick anything tells the real story. Choosing costs you the other thing, so it's harder to fake than just saying words.

Choices Tell The Truth

Revealed preference means figuring out what someone really values by watching what they choose, not by asking them. The big idea: when you pick one thing over others that were right there, you give up those others — that cost makes a choice hard to fake, while just saying "I prefer this" costs nothing. So if your words and your choices disagree, you trust the choices. From a bunch of choices across different options, you can work backward to figure out the person's real ranking of what they like. It can even break in a useful way: if someone picks A over B, B over C, but C over A, that loop tells you no clean ranking fits — and that's a clue too.

Choices Over Words

Revealed preference is the commitment to infer hidden valuations from observed choices rather than from what people say they want. The defining move is to treat the choice trace — which alternative someone picked when others were available — as the load-bearing evidence about what they value, and to treat verbal reports as a separate channel you're not obligated to trust. It rests on three parts: an epistemic asymmetry (choosing carries an opportunity cost, so it's harder to fake than reporting, which is free), a substitution claim (under sane behavioral assumptions, the chosen option carries at least as much information as a sincere report and strictly more than a distorted one), and an inversion procedure (from enough choices across varying menus you recover a preference structure that rationalizes them). That procedure has a known failure mode — intransitive cycles mean no rationalizing preference exists — and the failure is itself diagnostic.

 

Revealed preference is the structural commitment to infer latent valuations from observed choices rather than from solicited reports. The defining move is to treat the choice trace — the record of which alternative an agent picked when others were available — as the load-bearing evidence about what the agent values, and to treat any verbal report ("I prefer X to Y") as a separate and possibly inconsistent channel the analyst is not obligated to trust. The commitment has three tight components. First, an epistemic asymmetry: the choosing act is harder to fake than the reporting act, because choosing carries an opportunity cost while reporting does not. Second, a substitution claim: under reasonable behavioral axioms — consistency, weak transitivity, no spite or self-deception — the chosen alternative carries at least as much information as a sincere report and strictly more than a distorted one. Third, an inversion procedure: from a sufficient record of choices across varying menus, the analyst recovers a preference structure (or utility function, or value ranking) that rationalizes the choices. That inversion has a known failure mode — intransitive cycles imply no rationalizing preference exists — and the failure is itself diagnostic. The pattern is recognizable wherever an agent's acted behavior diverges from their stated behavior and the analyst must decide which to credit; naming it commits the analyst to credit the act over the statement, but with eyes open to the framing conditions under which that is sound and to the conditions under which the inversion breaks and the stated channel may be the better evidence.

Structural Signature

the agent whose valuations are sought — the menu of available alternatives — the chosen alternative — the stated alternative — the cost-of-signal asymmetry — the inversion procedure — the consistency axioms and failure diagnostic

A structure is revealed preference when each of the following holds:

• An agent. There is a chooser whose latent valuations are the object of inference.
• A menu of available alternatives. The agent faces a determinate set of options actually on offer; preferences are recoverable only over alternatives the agent confronted, so the menu is load-bearing.
• A chosen alternative. The agent picks one option when others were available — the choice trace, treated as the primary evidence.
• A stated alternative. A parallel, cheaper channel — what the agent reports they value — runs alongside the choice and may be inconsistent with it.
• The cost-of-signal asymmetry. Choosing carries an opportunity cost while reporting does not, so the enacted choice is harder to fake and credited over the spoken one.
• An inversion procedure. From a sufficient record of choices across varying menus, a preference structure (ranking or utility) is recovered that rationalizes the choices.
• Consistency axioms and a failure diagnostic. The inversion is valid under behavioral axioms (consistency, transitivity, no self-deception); intransitive cycles or framing-dependence signal that no rationalizing preference exists — a result that is itself informative and may make the stated channel the better evidence.

The components compose so that crediting the expensive act over the cheap report, under stated axioms, inverts a choice trace into a preference — with the inversion's known failure modes marking exactly when to distrust it.

What It Is Not

• Not preference itself. Revealed preference is an inference method — credit the choice trace — riding on the deeper notion preference. The recovered ranking is what one estimation procedure produced under its axioms, not the agent's valuation as such.
• Not a stated preference. It deliberately credits the enacted choice over the reported one; the two are separate channels with different cost-of-signal profiles, and the stated channel is exactly what revealed preference declines to trust by default.
• Not signaling. Signaling is a sender's deliberate costly act to communicate type; revealed preference is an observer's inference from choices the agent made for their own sake, not to signal.
• Not a time_preference_discounting_future claim. Discounting concerns how future payoffs are weighted; revealed preference is the broader method of inferring any valuation from choice, of which intertemporal choice is one application.
• Not regret or anticipated regret. Regret is an emotional/evaluative response to outcomes; revealed preference reads valuation from the choice itself, independent of how the agent feels about it afterward.
• Common misclassification. Dogmatically discarding the stated channel because behavior "contradicted" it — when the behavior was constrained, mis-instrumented, or the report carried information the choice could not (anticipated future preference, values not yet actable).

Broad Use

• Economics and consumer theory (origin): Samuelson's 1938 axiomatization, Houthakker's Strong Axiom, demand-system estimation from observed purchases, hedonic pricing.^[1]
• Behavioral analytics and UX: clickstream, dwell time, and A/B tests — what users do reveals preferences that surveys about the same page systematically misreport.^[2]
• Policy and political science: turnout patterns and "voting with one's feet" (Tiebout sorting, migration between jurisdictions) treated as more reliable than opinion polls on the same questions.^[3]
• Law and evidence: the doctrine that acts speak louder than words in contract interpretation (course of dealing, course of performance);^[4] inferring intent from conduct rather than post-hoc characterization.
• Sociology and family studies: time-use diaries and budget allocations as evidence of household priorities, often contradicting stated value rankings.
• Conservation and animal behavior: habitat-choice and optimal-foraging studies infer how an animal values food and safety from where it goes and when it stops, since the animal cannot report.^[5]
• Clinical research: the gap between stated treatment preferences and adherence behavior; shared-decision models that triangulate stated and revealed channels.
• Marketing: A/B tests and bandit experiments as machinery for eliciting preference from action, routinely outperforming focus-group elicitation.^[1]

Clarity

Naming the pattern makes vivid the otherwise-blurred line between two channels of information about the same agent: what the agent says they value and what the agent does under a real choice opportunity. The named distinction immediately raises a triage question for any preference-inference problem — which channel are we using, and why?

It also clarifies the cost-of-evidence asymmetry that drives the whole move. A reported preference is cheap (just talk); an enacted choice is expensive (a forgone alternative); and expensive signals are harder to fake. Practitioners routinely muddle the channels — running a survey when a choice experiment was available, or dismissing observed behavior because it contradicted what respondents said — and the named prime supplies a vocabulary for catching the mistake before it propagates into a conclusion.

A second clarification is that the move is not "ignore what people say" but "be explicit about which channel you are believing and under which behavioral axioms it is informative." Revealed preference's known failure modes — preference reversals across framings, intransitive cycles, status-quo dependence — are part of the prime's structural payload, not embarrassments to it. They tell the analyst precisely when the inversion procedure fails, and therefore when the stated channel, for all its distortions, may be the better one to credit.

Manages Complexity

The pattern compresses a large family of preference-inference problems — consumer demand, voting behavior, animal habitat choice, employee retention, patient adherence, evidence of intent in law — into a single diagnostic stance: prefer the choice trace to the verbal report, document the axioms under which the inversion is sound, and flag the conditions (framing effects, incomplete menus, social-desirability bias) under which the channels diverge in known ways.

That compression converts a wide design problem — how to elicit what an agent values — into a small intervention space. One can instrument the choice environment (make the menu visible, log the action, control for outside options); vary the menu to identify the preference, which is the standard revealed-preference experimental design; or triangulate channels by comparing the stated and revealed signals and diagnosing their divergence, rather than picking one and discarding the other. Each intervention is substrate-neutral: instrumenting a checkout flow, designing a foraging experiment, and reading a contract's course of performance are the same structural act of building a choice instrument instead of a survey instrument.

Abstract Reasoning

Recognizing revealed preference as a structural pattern enables reasoning about the two-channel structure of preference data: any preference-relevant question has at least two evidence streams, stated and revealed, each with a characteristic distortion profile, and naming the streams lets the analyst design collection around the cheaper and the costlier signal jointly. It connects to the cost-of-signal logic shared with signaling theory: signals the sender pays for carry information that cheap talk cannot, and revealed preference is the consumer-theory instance of an insight that recurs in biology and game theory.^[6]

It also sharpens reasoning about failure. The framing-dependence mode says that when a menu's presentation systematically shapes the choice — defaults, ordering, anchoring — the inversion from choices to preferences becomes ambiguous, and the response is to counterbalance framings rather than naively attribute the choice to "preference." The intransitivity diagnostic says that when revealed choices form cycles, no single preference rationalizes them, and that very fact is informative: it moves the analyst from "estimate the preference" to "model the multi-self or context-dependent agent." And the exit-versus-voice contrast generalizes the whole pattern: exit is a revealed preference, voice a stated one, and organizations that listen only to voice miss the information in exit — a structure that reappears in customer churn, citizen migration, employee retention, and student withdrawal.^[7]

Knowledge Transfer

The transfers are concrete and well-documented. The Samuelson–Houthakker discipline of inverting choices into preferences moved directly into recommender-system design, clickstream personalization, and attention measurement;^[2] the structural move is identical, with digital trace data standing in for market purchase data. The same inversion ported into animal behavior and ecology through optimal-foraging and habitat-choice theory, which adapted the consumer-theory machinery without inheriting the verbal-versus-acted framing, since animals cannot report. In law, the doctrine that conduct under a contract outweighs post-hoc characterization — codified in the course-of-dealing and course-of-performance rules — is the same move applied to inferring intent.^[4] Tiebout's "voting with one's feet" carried the logic into public finance, reading migration as revealed preference over bundles of local public goods.^[3]

What makes these genuine transfers is that the structural roles map cleanly across every substrate. The agent whose valuations are sought, the menu of alternatives actually faced (a load-bearing detail, since preferences are recovered only over alternatives the agent confronted), the chosen alternative (expensive, because of the forgone option), the stated alternative (the cheap parallel channel), the inversion procedure, the consistency axioms under which it is valid, and the failure-mode diagnostic that signals when it is not — these recur unchanged whether the substrate is a supermarket, a polling station, a foraging patch, or a clinic. Tellingly, the behavioral-economics critique of framing effects and preference reversals does not dissolve the prime; it refines it, because the failure modes are themselves cross-substrate — framing in marketing parallels framing in voting parallels framing in medical decision-making. A practitioner who carries the pattern into a new domain inherits not just a stance but a portable kit: build choice instruments rather than survey instruments, and use the intransitivity-and-framing diagnostics that survey methods cannot themselves detect.

Examples

Formal/abstract

Samuelson's Weak Axiom of Revealed Preference (WARP) is the prime's inversion procedure made into a testable invariant. The agent is a consumer facing prices and an income; the menu at any moment is the budget set — every bundle the consumer can afford. The chosen alternative is the bundle actually bought. WARP states the consistency axiom precisely: if bundle \(A\) is chosen when \(B\) was also affordable, then there must be no later situation in which \(B\) is chosen while \(A\) is affordable and not more expensive. In symbols, \(A\) revealed-preferred to \(B\) forbids \(B\) revealed-preferred to \(A\). The inversion procedure runs from here — given enough budget-set/choice pairs satisfying WARP (and its strong, transitive extension, SARP), a utility function exists that rationalizes every observed choice, recovered up to monotonic transformation.^[8] The prime's failure diagnostic is built in and concrete: a choice cycle (\(A\) over \(B\), \(B\) over \(C\), \(C\) over \(A\)) violates WARP, and the structural verdict is that no rationalizing preference exists — the data cannot have come from a single stable utility-maximizer, so the modeling move shifts from "estimate the preference" to "model a context-dependent or multi-self agent." The intervention this licenses: to identify a demand system you deliberately vary prices (vary the menu) and read the preference off the choice responses, rather than asking the consumer what they want.

Mapped back: the consumer, the budget set, the purchased bundle, the no-reversal axiom, and the cycle-detection test instantiate agent, menu, chosen alternative, consistency axiom, and failure diagnostic; WARP's violation is exactly the intransitivity diagnostic the prime names as itself informative.

Applied/industry

A product-analytics team, a public-finance economist, and a court are all crediting the expensive act over the cheap report. The analytics team has survey data saying users want a feature, but the revealed channel — clickstream, dwell time, an A/B test — shows they never use it; the prime's cost-of-signal asymmetry tells them to credit the enacted choice (which forgoes attention elsewhere) over the costless survey answer, and the intervention is structural: instrument the choice environment and vary the menu (the A/B test is literally revealed-preference experimental design). The public-finance economist applies Tiebout's "voting with one's feet": residents' migration between jurisdictions is a revealed preference over bundles of local taxes and public goods, read as more reliable than opinion polls on the same services — the exit-versus-voice contrast the prime generalizes, where exit (the costly move) carries information that voice (the cheap complaint) does not. A court interpreting a contract invokes the doctrine that acts speak louder than words: course-of-dealing and course-of-performance — what the parties actually did under the agreement — outweigh post-hoc characterizations of what they "intended," the same inversion from conduct to latent valuation. Each domain also inherits the failure caution: framing effects (a default-driven UI choice, a poll-question order) can make the revealed channel ambiguous, and the disciplined response is to counterbalance rather than naively attribute.

Mapped back: product analytics, public finance, and contract law are three genuine domains where the same roles operate — agent, menu, costly chosen act versus cheap stated report — and the prime's kit (build choice instruments, credit exit over voice, watch for framing) transfers intact, though the pattern carries its economics-origin framing with it.^[1]

Structural Tensions

T1 — Act versus Word (the channel the prime commits to, and its limit). Revealed preference credits the costly enacted choice over the cheap verbal report — but the commitment is conditional, not absolute. The characteristic failure mode is dogmatically dismissing the stated channel: discarding survey or testimony evidence because behavior "contradicted" it, when the behavior was constrained, mis-instrumented, or the report carried information the choice could not (anticipated future preference, values not yet actable). Diagnostic: ask whether the choosing act was genuinely freer and more informative than the report; the prime's own failure modes (framing, incomplete menus) mark exactly when the stated channel becomes the better evidence, so "credit the act" is a default with documented exceptions, not a law.

T2 — Menu-Dependence (preference is only recoverable over what was faced). The inversion recovers preferences solely over alternatives the agent actually confronted; an option absent from the menu reveals nothing. The failure mode is inferring a preference the agent never had the chance to express — reading "did not choose X" as "does not value X" when X was unavailable, unaffordable, or invisible. Diagnostic: ask whether the unchosen alternative was genuinely on the menu and within reach; if the choice set was constrained or incomplete, the revealed preference is conditional on that constraint, and generalizing it beyond the faced menu is unwarranted.

T3 — Transitivity versus Intransitive Cycles (when no preference exists). The inversion is valid only under consistency axioms; when revealed choices form cycles (A over B, B over C, C over A), no single rationalizing preference exists. The tension is that this is not noise to smooth away but a structural verdict. The failure mode is forcing a stable utility function onto cyclic data — estimating "the preference" of an agent that has none, masking a context-dependent or multi-self chooser. Diagnostic: test for cycles before fitting; if choices violate transitivity, the modeling move shifts from "estimate the preference" to "model a context-dependent agent," and a fitted utility is an artifact.

T4 — Stable Preference versus Framing-Constructed Choice (the inversion's ground gives way). Revealed preference assumes the choice expresses a pre-existing valuation the analyst inverts back to. But when defaults, ordering, and anchoring construct the choice, there is no stable prior preference to recover — the menu's presentation, not the agent's value, drove the act. The failure mode is attributing a framing-driven choice to "preference," then designing policy around a value the agent did not hold. Diagnostic: ask whether the choice survives counterbalanced framings; if reordering options or changing the default flips the choice, the inversion is ambiguous and the right response is to counterbalance presentations, not to read off a preference.

T5 — Static Trace versus Changing Preference (the temporal boundary). The inversion treats the choice record as evidence of a fixed valuation, but preferences drift, adapt, and are shaped by past choices (taste formation, addiction, learning). A trace aggregated over time can rationalize to a single preference that no version of the agent ever held. The failure mode is inferring one stable utility from choices spanning a period over which the agent's values changed — averaging a former self and a current self into a fictitious composite. Diagnostic: ask whether the agent's preferences plausibly shifted across the trace window; if choices early and late in the record reflect different valuations, the inversion must be time-localized, not pooled.

T6 — Revealed Preference versus Preference Itself (the framing boundary). Revealed preference is an inference method — credit the choice trace — riding on the deeper notion preference, its near-identical neighbour. The tension is conflating the evidence with the thing: treating the revealed ranking as the agent's preference, when it is only what one estimation procedure recovered under its axioms. The failure mode is reifying a methodological artifact as the agent's true value, hiding the economics-origin framing (rational consistent chooser, opportunity-cost-bearing acts) that the method imported and that may not hold. Diagnostic: ask whether "revealed preference" is naming the agent's valuation or the analyst's inference from one channel; if the axioms behind the inversion are doing unexamined work, the recovered ranking is a model output, not the preference itself.

Structural–Framed Character

Revealed preference sits at the upper edge of the mixed-framed band on the structural–framed spectrum, well past the midpoint. A genuine relational skeleton lies underneath — an agent, a menu, a chosen alternative, and an inversion that recovers a preference ordering rationalizing the choices — but it travels wrapped in consumer theory and a substantive epistemic claim, so naming it imports a frame rather than merely recognizing a pattern.

The two decisive diagnostics both read 1.0. Its institutional origin is squarely economic — Samuelson's 1938 construction within consumer demand theory — and to invoke it is to import that framework: the very phrase commits the analyst to a position (credit the act over the word) that carries epistemic and mildly normative weight about acts-versus-words, which is exactly an imported interpretive frame, not a neutral observation. Three diagnostics read 0.5. Its vocabulary travels with translation — menu, choice trace, rationalizing utility, consistency axioms are economic-theoretic terms that need restating in psychology or policy. And it is human-practice-bound in part: the pattern presupposes a choosing agent facing an opportunity cost, a relational role that does not run in indifferent physical substrates, though it is not confined to formal markets. The evaluative_weight reads 0.5 rather than 1.0 because the act-over-word commitment carries a real but not overwhelming normative charge. Two criteria at maximum and three half-steps sum to the 0.7 aggregate the frontmatter assigns — a relational inversion procedure carrying a heavy economics-and-epistemics frame, consistent with the mixed-framed label.

Substrate Independence

Revealed preference is a moderately substrate-independent prime — composite 3 / 5 on the substrate-independence scale. Its domain breadth is genuinely wide (scored 4): the infer-latent-value-from-the-choice-trace pattern recurs in economics and consumer theory (its origin), behavioral analytics and UX (clickstream, A/B tests), policy and political science ("voting with one's feet," turnout over polls), law and evidence (acts-speak-louder-than-words in contract interpretation), sociology (time-use diaries as priority evidence), conservation and animal behavior (habitat-choice and optimal-foraging studies, where the agent cannot report), clinical research (the stated-preference-versus-adherence gap), and marketing. What holds the structural abstraction down to 3 is how much the signature imports: it is squarely economics-bound (Samuelson 1938), the very phrase commits the analyst to a substantive epistemic position (credit the act over the word) carrying real normative charge, and its vocabulary (menu, choice trace, rationalizing utility, consistency axioms) needs translation when ported — so the pattern is recognized through a consumer-theory frame rather than as a value-neutral relational shape, and it presupposes a choosing agent bearing an opportunity cost rather than running in indifferent substrates. The transfer evidence is concrete (scored 4): the Samuelson–Houthakker inversion ported directly into recommender systems and clickstream personalization, into animal ecology via optimal-foraging theory (without the verbal-versus-acted framing, since animals cannot report), into public finance via Tiebout's "voting with one's feet," and into contract law via course-of-dealing doctrine — documented transfers where the roles (agent, menu, costly act, cheap report, inversion, failure diagnostics) map cleanly, and where the framing-effect critique refines rather than dissolves the prime. Strong, documented transfer within a heavily economics-framed band lifts the composite to 3; the inherited consumer-theory-and-epistemics frame is what caps it.

• Composite substrate independence — 3 / 5
• Domain breadth — 4 / 5
• Structural abstraction — 3 / 5
• Transfer evidence — 4 / 5

Relationships to Other Abstractions

Current abstraction Revealed Preference Prime

Parents (1) — more general patterns this builds on

• Revealed Preference presupposes Preference Prime

  Is emphatic: 'Revealed preference is an INFERENCE METHOD...

Children (6) — more specific cases that build on this

• Hedonic regression Domain-specific is a kind of Revealed Preference

  The proposed strict upward parent is prime:revealed_preference.

• Policy capturing Domain-specific is a kind of Revealed Preference

  The proposed strict upward parent is prime:revealed_preference.

• Willingness to pay Domain-specific is a kind of Revealed Preference

  The proposed strict upward parent is prime:revealed_preference.

▸ Show 3 more

Hierarchy path (1) — routes to 1 parentless root

• Revealed Preference → Preference

Neighborhood in Abstraction Space

Revealed Preference sits among the more crowded primes in the catalog (15^th percentile for distinctiveness): several abstractions describe nearly the same structure, so a description that fits it will tend to fit its neighbors too — transporting it usually means disambiguating within this family rather than landing on it exactly.

Family — Preference, Utility & Choice (18 primes)

Nearest neighbors

• Stated–Revealed Preference Gap — 0.79
• Stated Preference — 0.76
• Preference — 0.74
• Sunk Cost and Irreversible Commitment — 0.73
• Nirvana Fallacy — 0.72

Computed from structural-signature embeddings · 2026-09-10

Not to Be Confused With

Revealed preference must be distinguished from preference itself, its near-identical neighbour and the structure it is most dangerously conflated with. Preference is the underlying valuation — the agent's ordering over alternatives, whatever its source. Revealed preference is a specific inference method for recovering that valuation: credit the choice trace over the verbal report, invert the choices into a rationalizing ranking under stated behavioral axioms (consistency, transitivity, no self-deception). The recovered ranking is therefore a model output, not the agent's preference as such — it is what one estimation procedure produced, on one channel (acts, not words), under axioms that may or may not hold. The error is to reify the method's product as the thing itself: treating "the revealed preference" as the agent's true value when the agent may have intransitive choices (no rationalizing preference exists), framing-constructed choices (no stable prior preference to recover), or values the choice could not express (an unavailable option reveals nothing). The economics-origin framing — a rational, consistent, opportunity-cost-bearing chooser — rides along with the method and is exactly what may fail. The diagnostic is whether "revealed preference" names the agent's valuation or the analyst's inference from one channel; if the inversion's axioms are doing unexamined work, the recovered ranking is a fitted artifact, not the preference.

A second genuine confusion is with signaling, because both turn on the cost of an act carrying information. The distinction is in who pays and why. In signaling, a sender deliberately incurs a cost to communicate their type to an observer — the costly act is chosen because it transmits information (a credential, a warranty, a peacock's tail). In revealed preference, the agent chooses for their own sake, to get what they value, and the observer infers valuation from that choice as a byproduct — the agent is not trying to communicate anything. Both rely on the cost-of-signal logic (expensive acts are harder to fake than cheap talk), which is why they are kin, but the intentional structure is opposite: signaling is communication by the sender, revealed preference is inference about an agent who was not signaling. The error is to read a self-interested choice as a deliberate signal (over-attributing communicative intent — assuming a purchase was meant to send a message when it was just a purchase), or to read a genuine signal as mere revealed preference (missing that the act was strategically chosen to inform, so its information content is engineered rather than incidental).

These distinctions matter because each isolates a different question. Revealed-preference-versus-preference asks whether one is naming the agent's valuation or a method's output under axioms that may fail; revealed-preference-versus-signaling asks whether the costly act was inferred from (the agent chose for themselves) or deliberately sent (the agent chose to communicate). A practitioner who keeps them straight treats the recovered ranking as a model output rather than the preference itself, watches the axioms for unexamined work, and does not confuse an agent's self-interested choice with a sender's strategic signal.

Solution Archetypes

Solution archetypes in the catalog that build on this prime — directly (this prime is a source ingredient) or as a related prime.

Built directly on this prime (1)

• Revealed-Use Path Alignment: When people repeatedly cut their own path through a designed system, treat the trace as evidence and redesign the official path only after interpreting the cause, safety, legitimacy, and equity of the deviation.
  ▸ Mechanisms (12)

Also a related prime in 5 archetypes

• Access-Conditioned Bundle Decoupling: Prevent access leverage from forcing unwanted bundled acceptance by testing necessity, unbundling separable conditions, and preserving meaningful refusal, alternatives, or remedies.
• Coordination Scaffold Load Control: Keep the support machinery for coordination from becoming a second workload larger than the work it exists to coordinate.
• Cost-Asymmetric Preference Revelation Design: Reveal held preferences without forcing dangerous exposure by mapping expression costs, protecting low-risk signal channels, and translating aggregate private information into safer public choice and norm updates.
• Demand Curve Calibration and Response Design: Model how much of something is sought at different generalized costs, then use the calibrated response curve to guide allocation, pricing, capacity, and access decisions.
• Feasible-Alternative Comparator Calibration: Judge real options against reachable alternatives, not against perfection.

References

[1] Samuelson, Paul A. "A Note on the Pure Theory of Consumer's Behaviour". Economica, vol. 5, no. 17 (1938): 61–71. Originates revealed preference and the Weak Axiom (WARP), inferring preference from observed choice rather than introspected utility. registry ↩a ↩b ↩c

[2] De Corte, Kaat, John Cairns, and Richard Grieve. "Stated versus revealed preferences: An approach to reduce bias". Health Economics, vol. 30, no. 5 (2021): 1095–1123. Documents the systematic gap between what people say they prefer and what their behaviour reveals (the say–do / hypothetical-bias gap), and methods to reconcile the two channels. registry ↩a ↩b

[3] Tiebout, Charles M. "A Pure Theory of Local Expenditures". Journal of Political Economy, vol. 64, no. 5 (1956): 416–424. Models migration between jurisdictions as revealed preference over bundles of local public goods ('voting with one's feet'). registry ↩a ↩b

[4] American Law Institute and National Conference of Commissioners on Uniform State Laws. Uniform Commercial Code §§ 1-303, 2-208 (course of performance, course of dealing, usage of trade), first promulgated 1952. Codifies that the parties' actual conduct under an agreement evidences its meaning over post-hoc characterization. registry ↩a ↩b

[5] Stephens, David W., and John R. Krebs. Foraging Theory. Princeton: Princeton University Press, 1986. Infers how an animal values food and safety from observed habitat choice and patch-leaving (optimal-foraging) decisions, since the animal cannot report. registry ↩

[6] Spence, Michael. "Job Market Signaling". Quarterly Journal of Economics, vol. 87, no. 3 (1973): 355–374. Establishes the cost-of-signal logic that costly acts carry information cheap talk cannot — the kin insight to revealed preference. registry ↩

[7] Hirschman, Albert O. Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Cambridge: Harvard University Press, 1970. Contrasts exit (a costly revealed action) with voice (a stated complaint), the structure generalized across churn, migration, and withdrawal. registry ↩

[8] Houthakker, H. S. "Revealed Preference and the Utility Function". Economica, vol. 17, no. 66 (1950): 159–174. States the Strong Axiom of Revealed Preference (SARP) under which a transitive utility function rationalizing the observed choices is recoverable. registry ↩