Skip to content

Dictator Game

An experimental protocol where one participant unilaterally splits an endowment with a powerless recipient — by stripping away every strategic lever, it forces any positive transfer onto the non-strategic ledger, so its lattice of single-bit variants decomposes the sources of pro-social behaviour.

Core Idea

The dictator game is a minimal experimental protocol in which one participant — the dictator — receives a fixed monetary endowment and decides unilaterally how much, if any, to transfer to a second participant — the recipient — who has no power to accept, reject, or retaliate. Because the recipient's response cannot affect the dictator's outcome, the game-theoretic prediction under self-interested rationality is unambiguous: the dictator transfers nothing and keeps the entire endowment.

The empirical finding contradicts this prediction robustly across hundreds of replications. In the standard protocol, dictators transfer on average 20–30% of the endowment; the modal single allocation is often 50%. The protocol's value lies precisely in this contradiction: by stripping away all strategic structure — there is no punishment threat, no future interaction, no reputational stake in the canonical single-shot version — any positive transfer must be attributed to non-strategic motivation. Candidates include altruism (genuine concern for the recipient's payoff), inequity aversion (discomfort with unequal splits), social norms of fairness operating even without enforcement, and audience effects (the experimenter observes the dictator's choice, which alone may shift behavior).

Experimental variants are the diagnostic instrument. The double-blind protocol, in which neither the experimenter nor anyone else can identify the dictator's individual choice, substantially reduces transfers, implying that a significant fraction of standard-protocol giving is driven by observability rather than underlying other-regarding preferences. A taking variant — where the dictator can also take from the recipient's endowment — reveals that zero is not a natural reference point but a norm anchored by the experimenter's framing. Cross-cultural runs by Henrich and colleagues across fifteen small-scale societies found wide variation in dictator-game allocations, establishing that the observed generosity in standard Western laboratory populations reflects local institutional and market-integration norms rather than a universal human tendency. Together the paradigm and its variants form a lattice of controlled contrasts — dictator versus ultimatum game (with versus without punishment threat), dictator versus trust game (with versus without expectation of reciprocity), standard versus double-blind (with versus without audience) — that jointly decompose the sources of observed pro-social behavior.

Structural Signature

Sig role-phrases:

  • the endowment — a fixed windfall granted to one party for unilateral allocation
  • the dictator — the allocator who decides the split under no strategic constraint from the recipient
  • the powerless recipient — receives the chosen share with no power to accept, reject, retaliate, or reciprocate
  • the stripped baseline — the deliberate removal of rejection, repeat play, and (canonically) observability, which zeroes out every strategic motive by construction
  • the non-strategic residue — what the stripping guarantees: any positive transfer must be attributed to altruism, inequity aversion, a fairness norm, or audience effects, with nothing strategic left to confound it
  • the observed allocation distribution — the reading: mean transfer ~20–30%, mass at the 50% split, robustly contradicting the self-interest prediction of zero
  • the single-bit contrast lattice — the diagnostic apparatus completing the instrument: each neighbour or variant (vs. ultimatum = punishment threat; vs. trust = reciprocity; vs. double-blind = observability; taking option = framing; cross-cultural = local norm) differs by one feature, isolating one component
  • the context-laden number caveat — the limitation: the canonical figure is a local mix of motives, not a universal human constant, so only the contrasts, not any single allocation, license a component claim

What It Is Not

  • Not proof that humans are intrinsically generous. The canonical Western-laboratory figure (mean transfer 20–30%, modal 50%) is a context-laden measurement, not a universal human parameter. Cross-cultural runs found wide variation tracking market integration and local norms, and double-blind protocols showed much standard giving is observability-driven — so the number measures a local mix of motives, and only the lattice of contrasts licenses a claim about which component is present.
  • Not a strategic game. Its entire point is the removal of strategy: the recipient cannot accept, reject, retaliate, or reciprocate, so there is no punishment threat, future round, or reputational return to optimize against. The dictator's choice is parametric, not a best response — which is precisely what forces any positive transfer onto the non-strategic ledger.
  • Not a direct measure of altruism alone. A positive transfer is consistent with altruism, inequity aversion, an unenforced fairness norm, or the mere pull of being observed; the standard protocol cannot separate these. The instrument measures the non-strategic residue as a bundle, and only its variants (double-blind isolates the audience term, the taking option isolates framing) decompose it into named components.
  • Not the recipient's powerlessness being a flaw. The stripped, "impoverished" structure is not a limitation of the design but its calibration — removing every strategic lever is exactly what makes any transfer attributable to a non-strategic source. The feature that looks like a defect is the instrument's whole diagnostic value.
  • Not a real-world unilateral gift. A tip, a bequest, or a charitable transfer is a genuine act of giving, but calling it "a dictator game" borrows the name for something that was never the controlled protocol — dropping the experimenter-anchored zero, the fixed windfall, and the stripped baseline. The dictator game is laboratory hardware; the field act is an instance of the social-preference constructs it was built to isolate, not the paradigm itself.

Scope of Application

Because the dictator game is a controlled experimental protocol, not a causal mechanism, it applies wherever its precondition can be set up — a participant given an endowment and unilateral allocation power over a powerless recipient — and the fields below are real deployments of the identical instrument, read by the same logic of controlled subtraction. The boundary is instrument-reach versus over-reading: its canonical number is a local mix of motives, not a universal constant, and the other-regard it reveals travels onward under fairness / altruism / reciprocity, not as "the dictator game."

  • Experimental economics — the home laboratory, where it is one of the four or five core paradigms (with the ultimatum, trust, public-goods, and prisoner's-dilemma games) and its allocation distribution is a portable workhorse dataset, comparable across labs and disaggregable by treatment.
  • Behavioral game theory — where the modal positive transfer is a key empirical anchor for utility models augmented with inequity-aversion (Fehr-Schmidt, Bolton-Ockenfels), reciprocity, or warm-glow terms.
  • Cross-cultural anthropology / economics — Henrich et al.'s runs across fifteen small-scale societies, using dictator variants to test whether allocation norms track market integration rather than a universal human fairness norm.
  • Development economics and policy evaluation — dictator-style transfer protocols measuring social capital, group identity, and the effect of interventions on community-level other-regarding behavior.
  • Neuroeconomics — allocations correlated with medial-prefrontal, anterior-insula, and temporoparietal activation, informing models of how other-regarding preferences are neurally implemented.

Clarity

Naming the dictator game makes legible a distinction that richer games leave fatally entangled: strategic motives for giving — fear of punishment, anticipated reciprocity, reputational return — versus non-strategic ones — altruism, inequity aversion, fairness norms, the pull of being observed. The protocol's whole value is that by removing every strategic lever (no rejection, no future round, no retaliation), it forces any positive transfer to be attributed to a non-strategic source. This is what lets a behavioral economist read the ultimatum game correctly: generosity there could be genuine other-regard or merely fear of rejection, and only the dictator game's stripped baseline can tell those apart. The defining structural feature — the recipient is powerless — is not a limitation but precisely the instrument's calibration.

The sharper question it lets a researcher ask is not "are people pro-social?" but "which component of pro-sociality is producing this transfer, and would it survive removing the audience, the framing, or the cultural setting?" Each variant becomes a designed contrast that isolates one term: anonymity probes observability, the taking option probes whether zero is a true reference point or a framing artifact, cross-cultural runs probe whether the generosity is universal or locally institutional. And the game's deeper clarifying role is positional — it is one node in a lattice of controlled contrasts (dictator versus ultimatum isolates the punishment threat; dictator versus trust isolates expected reciprocity; standard versus double-blind isolates audience), so naming it crisply defines what every neighboring paradigm is measuring by difference from it.

Manages Complexity

Observed pro-social giving is a snarl of co-acting causes: altruism, inequity aversion, fairness norms, anticipated reciprocity, reputational return, fear of punishment, warm glow, and the mere pull of being watched all push on any real act of sharing at once, and in richer games they are confounded — a generous ultimatum offer could be genuine other-regard or naked fear of rejection, with no way to tell from the offer alone. The dictator game tames that snarl not by modeling the full motive mixture but by supplying a baseline that zeroes out an entire class of terms: strip away rejection, repeat play, and retaliation, and every strategic motive vanishes by construction, so whatever transfer remains is attributable to the non-strategic residue alone. That single subtraction collapses the dimensionality of the giving problem — from "which of a dozen entangled motives produced this?" to "the strategic terms are off; what is left?" The analyst tracks one allocation distribution (mean transfer, mass at the 50% split) and reads it as a direct measurement of non-strategic other-regard, no longer having to disentangle it from the strategic machinery that the protocol has removed.

The compression compounds because the paradigm is not one instrument but a lattice of controlled contrasts in which each comparison turns on a single bit, isolating one additive component of pro-sociality. Dictator versus ultimatum differs only in the punishment threat, so the difference in giving measures the strategic-fear term; dictator versus trust differs only in expected reciprocity, so the difference measures the reciprocity term; standard versus double-blind differs only in observability, so the difference measures the audience term; the taking variant differs only in where the experimenter places zero, so the shift measures the framing/norm-anchoring term; cross-cultural runs differ only in institutional setting, so the spread measures how much is local norm versus universal tendency. The high-dimensional question "what generates human generosity?" thereby reduces to a small set of pairwise differences, each reading off one parameter by holding everything else fixed — the analyst recovers the motive decomposition by tracking a handful of between-condition contrasts rather than fitting a full structural model of social preference to every new dataset.

Abstract Reasoning

Within experimental economics the paradigm licenses reasoning moves that all exploit the stripped baseline and the lattice of single-bit contrasts built around it.

Diagnostic — from a transfer made with every strategic lever removed, infer a non-strategic source, and read each treatment's shift as one component. The signature move attributes residual giving by elimination: because the recipient cannot accept, reject, retaliate, or reciprocate, the analyst reasons FROM "a positive transfer occurred in a game with no punishment threat, no future round, and no reputational return" TO "this transfer must come from a non-strategic motive — altruism, inequity aversion, a fairness norm, or the pull of being observed." A second diagnostic move reads each variant's effect as isolating one term: reasoning FROM "transfers fall sharply under the double-blind protocol" TO "a significant fraction of standard giving was driven by observability, not underlying other-regard"; FROM "allocations vary widely across small-scale societies" TO "the laboratory generosity reflects local institutional and market-integration norms rather than a universal human tendency"; FROM "behavior shifts when a taking option is added" TO "zero was a framing-anchored reference point, not a natural one." The reasoning is FROM an observed allocation (or its change between conditions) TO the specific non-strategic component responsible.

Interventionist — design each variant as a controlled subtraction that isolates one motive, and predict its effect. The prescribing move builds knowledge by adding or removing exactly one feature and predicting the resulting change: impose anonymity (double-blind) to remove the audience and predict a drop in transfers equal to the observability component; add a taking option to relocate the experimenter's zero and predict a shift measuring the norm-anchoring/framing term; run the same protocol across cultures to vary institutional setting and predict a spread measuring how much giving is local norm versus universal tendency. The analyst reasons FROM "change this single feature, holding all else fixed" TO "the difference in the allocation distribution measures the term that feature controls." A second interventionist move positions the dictator game as the deliberately stripped instrument against which richer games are read: reasoning FROM "the ultimatum game adds back a punishment threat" TO "the dictator-versus-ultimatum difference measures strategic fear, which the dictator baseline alone could not produce."

Boundary-drawing — separate strategic from non-strategic motives, and treat recipient powerlessness as calibration rather than limitation. A first boundary move holds the two motive classes apart, because richer games entangle them: generosity in the ultimatum game could be genuine other-regard or mere fear of rejection, so the analyst reasons FROM "remove every strategic lever" TO "only the dictator game's stripped baseline can attribute giving to a non-strategic source and thereby disambiguate the ultimatum offer." A second boundary move reclassifies the protocol's defining feature: reasoning FROM "the recipient is powerless" NOT to "this is an impoverished game" but TO "this is the instrument's calibration — the powerlessness is exactly what forces any transfer onto the non-strategic ledger." A third boundary move fixes the paradigm's place positionally: each neighbouring game is defined by difference from the dictator baseline (dictator versus ultimatum isolates the punishment threat; dictator versus trust isolates expected reciprocity; standard versus double-blind isolates audience), so the analyst reasons FROM "which single feature does this neighbour add" TO "what that neighbour therefore measures."

Predictive — self-interest forecasts zero and is robustly contradicted, while each single-bit contrast forecasts which term it will move. A forward move states the sharp prediction the protocol was built to test: under self-interested rationality the dictator is predicted to transfer nothing, and the analyst forecasts — confirmed across hundreds of replications — that real transfers cluster at 20–30% with a modal 50% split, so any positive giving is predicted to expose non-strategic motivation. A second predictive move forecasts the direction of each treatment effect before running it: removing the audience is predicted to reduce transfers, adding a taking option is predicted to reveal that zero is not the natural reference point, and moving to a non-market society is predicted to change the allocation norm — each contrast predicted to move exactly the one component it isolates while leaving the others fixed, so the full motive decomposition is forecast from a handful of pairwise differences rather than a fitted structural model.

Knowledge Transfer

The dictator game is a measurement instrument — a controlled experimental protocol — not a causal mechanism, so the "mechanism within / metaphor beyond" frame does not fit it; the relevant transfer is that of an instrument, which carries literally wherever its precondition (a participant who can be given an endowment and unilateral allocation power over a powerless recipient) can be set up, and the boundary to mark is instrument-reach versus over-reading. Within experimental and behavioral economics the protocol transfers as a paradigm: it is one of the four or five core instruments alongside the ultimatum, trust, public-goods, and prisoner's-dilemma games, and its allocation distribution is a portable workhorse dataset — comparable across labs, disaggregable by treatment, statistically tractable. The lattice of single-bit contrasts that gives it diagnostic power travels with it: dictator-versus-ultimatum to isolate the punishment threat, dictator-versus-trust to isolate expected reciprocity, standard-versus-double-blind to isolate observability, the taking variant to isolate norm-anchoring, cross-cultural runs to separate local institution from universal tendency. It reaches into development economics (transfer protocols measuring social capital and the effect of policy interventions), cross-cultural anthropology (Henrich et al.'s fifteen small-scale societies), and neuroeconomics (allocations correlated with medial-prefrontal, insula, and temporoparietal activation). In every one of these the instrument is the same instrument, deployed wherever its precondition holds and read by the same logic of controlled subtraction; the substrate differs, but the construct transfers without translation.

The over-reading to guard against is treating the instrument's reading as a substrate-free constant. The canonical Western-laboratory figure (mean transfer 20–30%, modal 50%) is not a universal human parameter: the cross-cultural runs established wide variation tracking market integration and local norms, and the double-blind result showed a large share of standard giving is observability-driven rather than underlying other-regard. So the over-read "the dictator game proves humans are intrinsically generous" mistakes a context-laden measurement for a fixed quantity — the instrument measures a local mix of motives, and only the lattice of contrasts, not any single number, licenses a claim about which component is present. A second over-read is purely verbal: calling any unilateral, unenforced gift — a tip, a bequest, a charitable transfer — "a dictator game" borrows the name for a real-world act that was never the controlled protocol, dropping the stripped baseline and the experimenter-anchored framing that make the paradigm an instrument at all. That is loose analogy, not the instrument traveling, and should be marked as such.

Where the findings generalize, they do so not as "the dictator game" but as instances of the constructs the instrument measures — and this is the genuinely portable layer. Unilateral allocation under no enforcement, giving without expected return, and observed fairness in the absence of audience recur across domains as instances of fairness, altruism, reciprocity, and the broader social-preferences / cooperation family; it is those patterns, recognized in field giving, institutional design, or non-laboratory settings, that carry the cross-domain lesson, while "dictator game" names the laboratory hardware that produced the evidence. The instrument transfers literally across the substrates where it can be run; its number must not be over-read as a constant; and the other-regard it reveals travels onward under the names of the social-preference primes it was built to isolate — the division of labor that Structural Core vs. Domain Accent makes precise below.

Examples

Canonical

The standard protocol, established as a control for the ultimatum game (Forsythe, Horowitz, Savin, and Sefton, 1994, building on an earlier version by Kahneman, Knetsch, and Thaler), runs like this. A dictator is handed a fixed endowment — say $10 — and told to divide it however they like with an anonymous recipient who has no power whatsoever: no ability to reject, retaliate, or interact again. Self-interested rationality gives an unambiguous prediction — keep all $10, transfer $0 — because nothing the recipient does can affect the dictator. Yet across hundreds of replications dictators give away, on average, roughly 20–30% of the stake, with a substantial mass of allocations at the equal 50/50 split. Since every strategic lever has been removed, that positive giving cannot be fear of punishment or hope of reciprocity; it must come from a non-strategic source.

Mapped back: The $10 is the endowment, the allocator is the dictator, and the anonymous no-power partner is the powerless recipient. Removing rejection and repeat play is the stripped baseline that zeroes out strategic motives, so the observed 20–30% mean and 50% mode — the observed allocation distribution contradicting the zero prediction — must be attributed to the non-strategic residue.

Applied / In Practice

Henrich and colleagues carried the instrument out of the Western laboratory and into fifteen small-scale societies — foragers, horticulturalists, pastoralists, and subsistence farmers across the Americas, Africa, Asia, and Oceania — running standardized allocation and bargaining experiments (dictator and ultimatum-type games). The result overturned the assumption that laboratory generosity reflects a fixed human trait: allocations varied enormously across societies, and the variation tracked structural features of each community, especially the degree of market integration and the payoffs to cooperation in daily life. Groups whose economies involved more anonymous market exchange tended toward more "fair" offers; more autarkic groups differed systematically. The experiments functioned as a controlled contrast in which the single varying feature was the institutional and cultural setting.

Mapped back: Each society's runs are the same instrument deployed on a new population, with only the institutional context changed — a use of the single-bit contrast lattice (cross-cultural setting as the varied feature). The wide spread of results is the context-laden number caveat made vivid: the Western 20–30% figure is a local mix of motives, not a universal constant, so market integration rather than a fixed non-strategic residue explains the differences.

Structural Tensions

T1: Powerlessness as calibration versus artificiality (clean attribution bought with ecological validity). The recipient's total powerlessness is the instrument's whole diagnostic value: by removing rejection, repeat play, and retaliation, it forces any positive transfer onto the non-strategic ledger, so the "impoverished" structure is calibration, not defect. But the same stripping that buys clean attribution is exactly what makes the setting maximally unnatural — real giving rarely occurs over a windfall one did not earn, to a wholly powerless stranger, with an experimenter-anchored zero. So the cleaner the isolation of non-strategic motive, the further the behavior sits from any context in which people actually give, and the harder it is to carry the finding back to the field without the strategic and relational structure the protocol deleted. The tension is that attribution-cleanliness and ecological validity trade off directly: the instrument's precision is its artificiality. Diagnostic: Is the stripped baseline being used to isolate a motive (its proper use), or is the laboratory allocation being read as if it described how people give in the strategically and relationally rich real world?

T2: Non-strategic residue versus its unseparated bundle (a class isolated, not a motive). The stripped baseline's achievement is real — it zeroes out every strategic motive, so any transfer must be non-strategic. But "non-strategic" is a class, not a motive: the residue still mixes altruism, inequity aversion, an unenforced fairness norm, and the pull of being observed, and the standard protocol cannot pull them apart. The double-blind result is decisive here, showing a large share of standard giving was observability-driven rather than underlying other-regard. So the headline "the dictator game measures altruism" over-reads a composite the instrument itself cannot decompose; only the lattice of variants can. The tension is that the subtraction which cleanly removes strategy leaves behind a bundle it has no power to resolve, and the single allocation number is routinely mistaken for a pure altruism reading. Diagnostic: Is the transfer being attributed to a specific motive (altruism), or only to the non-strategic class — and has a variant (double-blind, taking option) actually isolated the component being claimed?

T3: A portable number versus a context-laden reading (comparability invites over-reading). The allocation distribution's greatest asset is that it is a portable, comparable, statistically tractable workhorse — the same 20–30% mean and 50% mode reproduced across hundreds of labs. That very comparability tempts treating the figure as a human constant, a fixed measure of intrinsic generosity. But the double-blind protocol slashes it and Henrich's fifteen small-scale societies spread it enormously with market integration, establishing that the number is a local mix of motives, not a universal parameter. So the property that makes the instrument scientifically valuable (a stable, cross-lab reading) is exactly the property that misleads when the reading is reified. The tension is that a comparable number is both the instrument's power and the standing invitation to over-read a context-dependent measurement as a constant of human nature. Diagnostic: Is the canonical figure being used as a within-design contrast against a matched condition, or is a single context-laden allocation being asserted as a universal human tendency?

T4: The single-bit lattice versus the separability it assumes (do treatments really move one term?). The paradigm's diagnostic engine is the lattice of single-bit contrasts: dictator-versus-ultimatum isolates the punishment threat, standard-versus-double-blind isolates observability, the taking option isolates norm-anchoring — each difference read as measuring exactly one additive component. This works only if motives are separable and each treatment changes precisely one thing. But real manipulations are entangled: adding a taking option alters not just where zero sits but the whole choice architecture and the salient norm; changing the audience can shift framing and self-image at once; and components may interact non-additively rather than summing. The clean decomposition into terms rests on a separability assumption the treatments may quietly violate, so a between-condition difference credited to "the observability term" may carry a second, uncontrolled change. The tension is that the lattice's additive interpretation is as strong as the single-bit assumption is fragile. Diagnostic: Does the varied feature change exactly one motive term, or is the contrast confounding several changes that the "isolates one component" reading papers over?

T5: Autonomy versus reduction (a laboratory instrument or the social-preference primes it measures). The dictator game is not a causal mechanism but a controlled protocol, so it travels literally — not by analogy — wherever its precondition can be set up: development economics, cross-cultural anthropology, and neuroeconomics all run the same instrument, read by the same logic of controlled subtraction. Its proprietary hardware (the endowment, the stripped baseline, the allocation distribution, the single-bit lattice) is what carries there. But reduction bites at two edges: calling a real-world tip, bequest, or charitable gift "a dictator game" borrows the name for an act that was never the controlled protocol; and the findings generalize not as "the dictator game" but as instances of the constructs it isolates — fairness, altruism, reciprocity, the broader cooperation/social-preferences family. Those primes, recognized in field giving or institutional design, carry the cross-domain lesson, while "dictator game" names the laboratory apparatus that produced the evidence. The tension is between an instrument that transfers literally where runnable and the social-preference primes that carry its lessons everywhere else. Diagnostic: Resolve toward the primes (fairness, altruism, reciprocity, cooperation) when describing unilateral unenforced giving in the field; toward the named protocol only when the actual controlled instrument — endowment, powerless recipient, stripped baseline — is being run.

Structural–Framed Character

The dictator game sits at mixed on the structural–framed spectrum, with an unusual profile because it is not a causal mechanism but a measurement instrument — a designed laboratory protocol — so the criteria read partly as those of an artifact. On evaluative_weight it is structural: the protocol praises and blames nothing; it is a neutral apparatus for isolating a component of behavior, and the entry is careful that its own headline number licenses no "humans are generous" verdict. On human_practice_bound it is framed: the dictator game is laboratory hardware constituted by the practice of experimental economics — the endowment, the powerless recipient, the experimenter-anchored zero, the stripped baseline are all features of a designed research procedure; remove the experiment and there is no dictator game, only real-world giving that is not the protocol (a tip or bequest, the entry insists, is not "a dictator game"). Institutional_origin points the same way — the protocol, its variants, and its single-bit contrast lattice are artifacts of a research discipline (Forsythe et al., McCracken-era experimental design), not a structure a nature observer would find running observer-free. Vocab_travels is bimodal in the instrument's characteristic way: the apparatus transfers literally wherever its precondition can be set up (development economics, anthropology, neuroeconomics run the same instrument), but the name does not travel to describe field giving, and its canonical number is a context-laden local mix, not a portable constant. Import_vs_recognize: within the substrates where it can be run the instrument is deployed as the same instrument (recognition, not analogy), while its findings generalize only under the names of the constructs it isolates.

The genuinely portable content is not a skeleton the protocol is but the social-preference constructs it was built to reveal — unilateral unenforced giving, giving without expected return, observed fairness absent an audience — which travel as instances of the umbrella primes fairness, altruism, reciprocity, and the broader cooperation / social-preferences family (the instrument's own method, controlled subtraction that strips confounds to isolate one component, is general experimental-design logic rather than a proprietary structure). Those primes carry the cross-domain lesson; the dictator game's distinctive cargo — the endowment, the stripped baseline, the allocation distribution, the single-bit lattice — is the domain accent that stays in the laboratory and keeps the entry domain-specific. The cross-domain reach belongs to the social-preference primes, not to "the dictator game." Its character: an evaluatively neutral, discipline-built measurement protocol that transfers literally wherever it can be run but whose name and number stay home, structural only in the social-preference constructs it isolates and mixed by the laboratory apparatus that constitutes it.

Structural Core vs. Domain Accent

This section decides why the dictator game is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity — there is no separate section for that. The dictator game is unusual: it is not a causal mechanism but a measurement instrument, so the decisive question is not "framed or structural" but what the instrument shares with the wider catalog versus what stays laboratory hardware.

What is skeletal (could lift toward a cross-domain prime). Two things could lift, and they must be named separately. First, the behavior the instrument reveals: unilateral allocation under no enforcement, giving with no expected return, fairness observed without an audience — a thin relational structure of other-regard that survives stripping the laboratory away, and that recurs across domains as instances of fairness, altruism, reciprocity, and the broader cooperation / social-preferences family. Second, and distinct, the instrument's method: controlled subtraction — remove exactly one confound and read the between-condition difference as one isolated component, assembling a lattice of single-bit contrasts. That method is genuinely portable, but it is general experimental-design logic, not a structure proprietary to this protocol. Both are cores the dictator game shares — the first with the social-preference primes, the second with all controlled experimentation — and neither is what makes the dictator game distinctive.

What is domain-bound. Almost all the distinctive content is experimental-economics hardware and none of it survives extraction intact: the fixed endowment granted for unilateral allocation; the powerless recipient who cannot accept, reject, retaliate, or reciprocate; the stripped baseline that zeroes out every strategic motive by construction; the experimenter-anchored zero; the allocation distribution (mean transfer 20–30%, mass at the 50% split) as the workhorse reading; and the specific single-bit contrast lattice (dictator-vs-ultimatum for the punishment threat, dictator-vs-trust for reciprocity, standard-vs-double-blind for observability, the taking variant for framing, cross-cultural runs for local norm). These are the worked apparatus, the protocol, and the canonical datasets (Forsythe et al.'s $10 split; Henrich et al.'s fifteen small-scale societies) the discipline actually runs — all specific to a controlled laboratory setup with an experimenter, an endowment, and a designed contrast. The decisive test: remove the controlled protocol — a real tip, bequest, or charitable gift — and the endowment, the stripped baseline, and the experimenter-anchored zero have no referent; what remains is a genuine act of giving that is an instance of the social-preference constructs, not the paradigm, so calling it "a dictator game" borrows the name for something that was never the instrument.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose cross-domain transfer is recognition of the same mechanism, not analogy. The dictator game's transfer is bimodal in an instrument-specific way. As hardware it transfers literally wherever its precondition can be set up — a participant given an endowment and unilateral allocation power over a powerless recipient — so experimental economics, behavioral game theory, development economics, cross-cultural anthropology, and neuroeconomics all run the same instrument, read by the same logic of controlled subtraction; this is recognition of the identical protocol, not analogy, but it is bounded to substrates where the protocol can actually be run. Beyond that, two over-reads mark the limit: treating the instrument's number as a substrate-free human constant (the canonical figure is a local mix of motives, overturned by the double-blind drop and the cross-cultural spread), and borrowing the name for field giving that was never the controlled protocol. And when the bare structural lesson is needed cross-domain — other-regarding behavior in the absence of enforcement — it is carried not by "the dictator game" but, in more general form, by the primes it was built to isolate: unenforced giving is altruism, discomfort with unequal splits is fairness, giving-with-return is reciprocity, and the whole family is cooperation / social preferences. The cross-domain reach belongs to those parents; "the dictator game," as named, is the laboratory apparatus — endowment, stripped baseline, allocation distribution, single-bit lattice — that produced the evidence and stays home.

Relationships to Other Abstractions

Local relationship map for Dictator GameParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Dictator GameDOMAINPrime abstraction: Other-Regarding Preferences — is a decomposition ofOther-RegardingPreferencesPRIME

Current abstraction Dictator Game Domain-specific

Parents (1) — more general patterns this builds on

  • Dictator Game is a decomposition of Other-Regarding Preferences Prime

    Removing the unilateral-allocation laboratory apparatus leaves the cross-payoff preference that any positive transfer reveals once strategy is stripped away.

Hierarchy path (1) — routes to 1 parentless root

Not to Be Confused With

  • Ultimatum game. The nearest neighbour: identical except the recipient can reject the offer, in which case both get nothing — restoring a punishment threat. The dictator game is precisely the ultimatum game with that lever removed, and the dictator-versus-ultimatum difference measures strategic fear of rejection. A generous ultimatum offer may be self-protective; a generous dictator transfer cannot be. Tell: can the recipient reject and impose a cost (ultimatum), or are they wholly powerless (dictator)?

  • Trust game. A neighbour that adds reciprocity: the first mover sends money that is multiplied en route, and the second mover chooses how much to return, so sending can be a self-interested investment. The dictator game removes any expectation of return, so its transfer cannot be strategic reciprocity. Dictator-versus-trust isolates the reciprocity term. Tell: can the transfer be repaid by the other party (trust game), or is there no channel for return at all (dictator)?

  • Public goods game. A multi-player paradigm where each participant contributes to a shared pot that is multiplied and split, pitting private incentive against group benefit — a study of cooperation under free-riding temptation. The dictator game is a two-party, non-strategic allocation with no such tension. Tell: is the structure many contributors facing a free-rider incentive (public goods), or one allocator over a powerless recipient with no strategic incentive at all (dictator)?

  • Real-world unilateral gift (tip, bequest, charity). A genuine act of unenforced giving in the field. It resembles a dictator allocation and is an instance of the social-preference constructs the game isolates, but it is not the controlled protocol — it lacks the fixed windfall endowment, the experimenter-anchored zero, and the stripped baseline. Calling it "a dictator game" borrows the laboratory name for something that was never the instrument. Tell: is this a designed experimental protocol with an endowment and controlled baseline (the dictator game), or an act of giving in the world it was built to study (a social-preference instance)?

  • The social-preference primes it measures (fairness, altruism, reciprocity, cooperation). The substrate-neutral constructs the instrument was built to reveal — other-regard, unenforced giving, discomfort with inequality. These are what carry the game's findings into field giving and institutional design; the dictator game is the laboratory apparatus that produces evidence about them, not the constructs themselves. Tell: are you describing the behavior of caring about others' payoffs in general (these primes), or the specific controlled protocol that isolates it (the dictator game)? (Treated more fully in earlier sections.)

Neighborhood in Abstraction Space

Dictator Game sits in a crowded region of the domain-specific corpus (36th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.

Family — Strategic Traps & Market Structure (15 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12