Fact-checked by Grok 2 weeks ago

Dictator game

The Dictator game is a foundational experimental paradigm in behavioral and experimental economics designed to investigate social preferences, including altruism, fairness, and inequity aversion, by presenting participants with a decision under conditions of unilateral power.^[1] In its standard form, the game involves two anonymous players: the "dictator," who receives a fixed endowment (typically a sum of money) and unilaterally decides how to divide it between themselves and the "recipient," who has no influence over the allocation and simply accepts whatever is offered, with the dictator retaining the rest.^[2] This setup eliminates strategic considerations present in related games like the Ultimatum game, isolating the dictator's intrinsic motivations for sharing.^[3] The game originated in a 1986 study by psychologists Daniel Kahneman, Jack Knetsch, and economist Richard Thaler, who used a hypothetical version to explore how fairness norms constrain self-interested behavior in market-like scenarios, such as a store manager deciding whether to impose a price surcharge during a snowstorm. It was formalized as an incentivized laboratory experiment in 1994 by economists Robert Forsythe, Joel Horowitz, N. E. Savin, and Martin Sefton, who contrasted it with the Ultimatum game to demonstrate that positive transfers in the Dictator game reflect genuine other-regarding preferences rather than reciprocity or punishment fears. Standard economic theory, assuming rational self-interest, predicts that dictators will allocate nothing to the recipient and keep the entire endowment, yet empirical results consistently show otherwise: across hundreds of studies, dictators typically offer 20-30% of the endowment on average, with many giving exactly 50% to achieve equality.^[2] These findings challenge neoclassical assumptions and highlight the role of social norms in economic decision-making.^[3] Since its inception, the Dictator game has become a "workhorse" tool in experimental economics, applied in over 100 published experiments to examine influences on giving behavior, such as anonymity, stakes size, cultural differences, gender, and framing effects.^[2] For instance, higher anonymity or larger endowments often reduce giving, while real-world contexts (e.g., charitable appeals) can increase it, revealing how situational factors moderate prosociality.^[1] Variants, including sequential, multiplayer, or field-based versions, extend its utility to study topics like social influence, inequality, and policy design. Meta-analyses confirm robust patterns, with giving rates stable over time but varying by methodological details like payment incentives.^[2] Overall, the game underscores that human economic behavior is shaped not only by material incentives but by ethical considerations and expectations of fairness.

Introduction and History

Definition and Basic Rules

The dictator game is a fundamental experimental paradigm in behavioral economics designed to isolate and measure individuals' unilateral decisions regarding resource allocation in the absence of strategic retaliation or negotiation. In this two-player game, participants are assigned distinct roles: one as the dictator and the other as the recipient, with the interaction typically conducted anonymously to minimize social pressures. The dictator receives a fixed endowment from the experimenter, commonly $10 or an equivalent amount in experimental currency, and must decide how to divide it between themselves and the recipient. The recipient plays a passive role, possessing no opportunity to accept, reject, or influence the proposed allocation, which ensures that the dictator's choice reflects pure unilateral generosity or self-interest without fear of reprisal. Payoffs are straightforward: if the endowment is denoted as E, the dictator chooses to keep an amount X where $0 \leq X \leq E, retaining X for themselves and allocating the remainder E - X to the recipient. This binary division highlights the dictator's discretion over the entire sum, often presented through discrete choice options (e.g., keeping all, splitting equally, or giving all) to facilitate decision-making in laboratory settings. The primary objective of the dictator game is to assess prosocial behavior, such as altruism or adherence to fairness norms, by observing how much (if any) of the endowment the dictator voluntarily transfers, free from the strategic elements present in related paradigms like the ultimatum game where rejection is possible. As part of broader experimental economics games, it provides a controlled environment to probe deviations from pure self-interest, revealing insights into intrinsic motivations without external incentives or punishments.

Historical Development

The dictator game was first introduced in 1986 by Daniel Kahneman, Jack L. Knetsch, and Richard H. Thaler as part of a series of experiments testing fairness perceptions and the assumptions of economic theory, particularly in the context of market entitlements and prospect theory applications. In their initial setup, participants were given windfall endowments and asked to allocate portions to anonymous recipients, revealing deviations from purely self-interested behavior. The game was formalized and distinguished from related bargaining protocols in 1994 by Robert Forsythe, Joel L. Horowitz, N. E. Savin, and Martin Sefton, who designed it explicitly to isolate intrinsic fairness motives by removing the recipient's rejection power, thus eliminating strategic concerns present in the ultimatum game.^[4] This version, often termed the "dictator game" to emphasize the unilateral decision-making, became the standard experimental tool for measuring altruistic or egalitarian preferences without confounding reciprocity.^[4] During the 1990s, the dictator game gained early adoption within behavioral economics, frequently paired with the ultimatum game—introduced by Werner Güth, Rolf Schmittberger, and Bernd Schwarze in 1982—to contrast pure altruism against strategic fairness. This integration helped establish the game as a cornerstone for probing deviations from rational choice models, with studies accumulating rapidly to explore social norms and other-regarding behavior.^[5] Key milestones in the 2000s included expansions into developmental psychology, such as Harbaugh, Krause, and Berry's 2003 study on bargaining behavior in children aged seven to eighteen, which used the dictator game to track the emergence of fairness considerations over time. By 2011, Christoph Engel's meta-analysis synthesized evidence from over 100 published dictator game experiments, highlighting consistent patterns of positive giving and influencing subsequent methodological refinements. Subsequent meta-analyses, such as Krawczyk (2022), have confirmed and extended these patterns with data from additional lab experiments.^[6] In the post-2020 period, the dictator game has seen growth in digital implementations and cross-disciplinary applications, including online platforms for broader participant recruitment and integrations with fields like neuroscience and public policy.^[7]

Theoretical Context

Relation to Other Economic Games

The dictator game is closely related to the ultimatum game, serving as a simplified variant that removes the recipient's ability to reject the offer, thereby isolating pure altruism or fairness preferences from strategic considerations of punishment or reciprocity.^[4] In the ultimatum game, proposers typically offer around 40% of the endowment due to the risk of rejection, whereas dictator game allocations average about 28-30%, highlighting how the absence of rejection power reduces offers and underscores the role of strategic motives in the former.^[1] This design was specifically developed by Forsythe et al. (1994) to "purify" ultimatum game results by eliminating fairness-driven rejections, allowing researchers to test whether observed generosity stems from intrinsic other-regarding behavior rather than fear of retaliation.^[4] The dictator game also shares conceptual overlaps with cooperation-testing paradigms like the prisoner's dilemma and public goods games, all of which probe prosocial tendencies and the tension between self-interest and collective benefit.^[1] However, unlike the simultaneous, symmetric decisions in the prisoner's dilemma—where mutual defection is the Nash equilibrium and cooperation rates hover around 47%—or the group-based contributions in public goods games, the dictator game features unilateral, one-sided allocation without interdependent strategies or multiplication of contributions, emphasizing individual exploitation potential in a constant-sum setting.^[1] Within game theory taxonomy, the dictator game functions as a non-strategic variant of bargaining games, contrasting with sequential reciprocity models like the trust game, where the first mover's investment can be reciprocated or exploited by the second, leading to trustor sending rates of about 50% and trustee returns of 37%.^[1] This positions the dictator game as a baseline for measuring unconditional giving, free from the uncertainty and mutual dependence that characterize trust or bargaining dynamics in other games.^[8]

Predictions from Economic Theory

In standard economic theory, the dictator game is analyzed under the assumption of Homo economicus, where agents are purely self-interested and maximize their own material payoff without regard for others. In this framework, the dictator is predicted to keep the entire endowment E, allocating zero to the recipient (X = E, recipient receives 0), as any positive transfer reduces the dictator's payoff without benefit. From a game-theoretic perspective, the dictator game has a unique subgame perfect Nash equilibrium in which the dictator takes all of the endowment. Since the recipient has no decision right or recourse after the allocation, there is no strategic interaction, and the dictator's dominant strategy is full self-allocation, consistent with non-cooperative game theory.^[9] Neoclassical utility maximization formalizes this prediction, where the dictator's utility depends solely on their own consumption: U(X) = X, with no altruism or other-regarding terms. The optimization problem is thus \max U(X) subject to $0 \leq X \leq E, yielding the corner solution X = E. Extensions incorporating impure altruism, such as models where the dictator derives utility from both their own payoff and the recipient's (e.g., U = u(X) + v(E - X)), allow for positive giving if v > 0. However, the baseline neoclassical prediction remains minimal or zero transfers unless altruism is explicitly parameterized, providing a benchmark for deviations.^[10]

Experimental Implementation

Standard Protocol

The standard protocol for conducting the Dictator game in laboratory settings emphasizes reproducibility, anonymity, and minimal interference to isolate the dictator's unilateral decision-making. Participants are typically recruited from university subject pools, such as undergraduate students, and paired anonymously with another participant from the same pool.^[11] Roles—dictator (or allocator) and recipient—are randomly assigned to ensure no prior knowledge of assignment influences behavior, with equal numbers of each role to facilitate matching.^[12] Sessions are conducted in controlled environments like behavioral labs.^[1] Instructions are delivered either orally by the experimenter or in written form distributed to participants, using neutral language to avoid priming fairness or altruism (e.g., referring to players as "Allocator" and "Recipient" rather than "Dictator").^[12] The allocator receives an endowment, commonly €10 or $10, and must decide how much (if any) to transfer to the recipient, who has no decision rights and receives only the allocated amount.^[12]^[11] Choices are made via a simple response sheet, computer interface, or multiple-choice form allowing continuous allocations from 0 to the full endowment in increments (e.g., €0.50 or $1).^[12] Communication between participants is strictly prohibited, and the one-shot nature of the game ensures no repeated interactions.^[11] Anonymity is a core feature, implemented through double-blind procedures where neither the recipient nor the experimenter can link decisions to specific individuals, reducing social pressure or experimenter demand effects.^[11] Payments are disbursed privately at the session's end, often in cash or via vouchers, based solely on the game's outcome to incentivize realistic choices.^[12] Data collection focuses on the allocator's transfer amount, recorded anonymously for analysis, with recipients simply noting their received payoff.^[1] Following the game, a brief debriefing session explains the study's purpose—typically to investigate decision-making under unilateral control—while adhering to ethical guidelines like those from the American Psychological Association, which discourage deception unless necessary.^[12] Participants are informed that their data will be used anonymously for research, and any questions are addressed without revealing others' choices.

Methodological Variations

Researchers have introduced various methodological adjustments to the standard Dictator game protocol to address specific research objectives, such as enhancing incentive compatibility or adapting to different participant pools. These variations maintain the core structure—one player (the dictator) unilaterally allocates an endowment between themselves and a recipient—while altering procedural elements to isolate variables of interest.^[13] Anonymity levels represent a primary variation, with protocols ranging from full anonymity to partial revelation. In full anonymity setups, often implemented as double-blind procedures, neither the experimenter nor the recipient can identify the dictator's decision or identity, typically achieved through private decision-making in isolated booths and anonymous payment mechanisms like sealed envelopes or hidden ballot boxes.^[14] Partial anonymity, or single-blind protocols, conceals the dictator's choice from the recipient but allows the experimenter to monitor decisions via computer interfaces or coded responses.^[13] These distinctions, pioneered in seminal work by Hoffman et al. (1994), enable investigations into social image concerns without altering the game's fundamental payoff structure.^[14] Endowment forms also vary to manipulate perceived ownership or salience. Dictators may receive windfall endowments as straightforward cash payments or tokens exchangeable for money, with common amounts ranging from $5 to $20 to balance feasibility and psychological impact.^[13] Earned endowments, where participants complete tasks like quizzes or puzzles to obtain the pie, contrast with windfalls to probe entitlement effects, as initially explored in early experimental designs.^[13] Token systems, used in both lab and field contexts, facilitate precise control over allocation increments, such as in $0.50 units. Delivery methods adapt the game to diverse settings, including in-person laboratory sessions, online platforms, and field environments. In-person implementations, standard since Forsythe et al. (1994), involve participants gathered in controlled lab spaces using paper forms or computers for decisions, ensuring high compliance through direct oversight. Online delivery, increasingly common via platforms like Amazon Mechanical Turk (MTurk) or Prolific, recruits broader, non-student samples through web-based interfaces with automated randomization and payment via digital wallets, though requiring comprehension checks to mitigate distractions. Field settings deploy the game in natural contexts, such as among workers in real workplaces, using portable materials like printed slips to capture ecologically valid responses. Recent meta-analyses (as of 2025) indicate that non-standard protocols, such as those where all participants act as both dictators and recipients, can lead to more efficiency-oriented giving behavior compared to the traditional half-role design.^[15] Incentive compatibility adjustments distinguish real monetary stakes from hypothetical scenarios. Real-stakes protocols provide actual payoffs drawn from the endowment, upholding the one-shot nature typical of the game to mimic binding choices, as emphasized in foundational implementations. Hypothetical variants present allocations without payment, useful for large-scale surveys, while rare repeated iterations—where the same pairs interact multiple times—test consistency, often with role anonymity preserved across rounds.^[13] Random payment mechanisms, paying only a subset of participants, further enhance realism in repeated contexts.^[13] To mitigate order effects in multi-game sessions, where the Dictator game is paired with others like the Ultimatum game, experimenters employ randomization of presentation order across subjects or sessions. This between-subjects or within-session shuffling prevents priming from prior tasks, a practice standardized in experimental economics to ensure unbiased decision-making.

Empirical Findings

In standard dictator game experiments, dictators typically allocate 20-30% of their endowment to the recipient, substantially deviating from the self-interested prediction of 0% giving under neoclassical economic theory. A meta-analysis encompassing 129 papers, 616 treatments, and 41,433 observations reported an average giving rate of 28.3% of the endowment. In these settings, approximately 36% of dictators give nothing, while 17% opt for an equal split, with giving amounts often concentrated in discrete choices like 0%, 50%, or 100% when the action space is limited. For example, in laboratory experiments with a 10-unit endowment, mean giving frequently falls around 2.5-3 units.^[16] Developmental patterns reveal baseline sharing from an early age, with children aged 3-5 years allocating around 20% of resources in dictator games. Giving increases through childhood and adolescence toward more egalitarian allocations, before stabilizing in adulthood at rates similar to the overall adult average of 20-30%. These patterns hold across repeated measures and various endowment sizes, underscoring the robustness of prosocial tendencies in non-strategic allocation tasks.^[17]^[18] The core findings demonstrate consistency in baseline behavior across diverse small-scale samples, though giving rates exhibit some variation; for instance, rates are often higher in Western laboratory contexts compared to certain non-Western field settings. This empirical regularity, with dictators consistently exhibiting altruism beyond theoretical expectations, has been replicated in over 600 experimental treatments worldwide.^[16]

Cross-Cultural and Demographic Patterns

Cross-cultural studies of the Dictator game reveal systematic variations in sharing behavior linked to cultural orientations, with lower offers typically observed in individualistic societies compared to more collectivist ones. For instance, in the United States, an individualistic culture, university students offer around 20-25% of the endowment on average, reflecting a stronger emphasis on self-interest.^[19] In contrast, among small-scale societies with collectivist norms, such as the Tsimane in Bolivia or the Orma in Kenya, mean offers reach 30-32%, suggesting greater influence of group-oriented fairness norms.^[19] Cross-cultural datasets confirm these patterns, with offers varying due to local institutions and market integration. Overall baseline sharing in Dictator games averages 25-30% across studies. Gender differences in baseline Dictator game behavior consistently show women as more generous than men. A 2025 laboratory study with 1,161 participants found women transferring 35% of their endowment on average, compared to 25% for men, a gap of 10 percentage points that held after controlling for age, cognitive ability, and personality.^[20] This pattern aligns with evidence from economic games indicating women give more than men, with 92% of women making non-zero offers versus 75% of men.^[20] The difference is attributed to inherent prosocial tendencies rather than strategic factors in anonymous setups. Age and developmental stage also influence sharing, with giving increasing from childhood to young adulthood and stabilizing thereafter. Among children aged 3-10, egalitarian sharing averages around 20% in mini-Dictator games, rising moderately to 30% by early adolescence due to growing fairness concerns and theory of mind development.^[18] In adulthood, meta-analyses indicate older participants (over 40) give 5-15% more than younger ones (under 30), with effects strongest in non-student samples where giving plateaus after young adulthood.^[13] Among other demographics, higher education levels positively correlate with greater giving in baseline setups, as more educated individuals tend to offer 5-10% more, possibly reflecting enhanced empathy or social norm awareness.^[21] In contrast, income shows no strong effects on sharing, with meta-studies finding negligible differences across income brackets in standard anonymous conditions.^[22]

Influences on Behavior

Effects of Stakes and Anonymity

The effect of stakes on dictator behavior has been extensively examined through controlled experiments, revealing that higher endowments typically lead to a lower proportion of giving, though absolute transfers may increase modestly. In one study, dictators offered an average of 24.1% of a $10 endowment, dropping to 15.2% of a $100 endowment and further to 10.8% of a $10,000 endowment, indicating a significant negative relationship between stake size and proportional generosity (F(2, 199) = 69.24, p < 0.001).^[23] This pattern aligns with broader meta-analytic evidence showing that increased endowments reduce the share allocated to recipients across multiple dictator game implementations. Absolute giving amounts, however, often remain comparable or rise slightly with stakes, suggesting that while the relative cost of altruism diminishes, baseline sharing tendencies persist without proportional scaling. Anonymity levels also systematically influence allocations, with greater anonymity reducing giving due to diminished concerns over social image and reputation. For instance, when dictators' identities were concealed (no names revealed), average offers were 18.3% of the endowment, compared to 27.2% when family names were disclosed to recipients, representing a roughly 50% increase in generosity under reduced anonymity.^[24] This drop of approximately 9 percentage points under full anonymity highlights how observability fosters prosocial behavior by activating norms of fairness and avoiding disapproval.^[24] Mechanisms underlying this effect include lowered accountability and reduced anticipation of judgment, which weaken intrinsic motivations for equitable division.^[24] The impacts of stakes and anonymity often interact, with anonymity exerting a stronger influence at lower endowments where social pressures are more salient relative to material costs. At small stakes, the anonymity-induced reduction in giving can exceed 10 percentage points, as image concerns dominate decisions, whereas at high stakes, economic incentives overshadow observability effects, leading to proportionally smaller drops (around 5%). This interaction underscores how situational factors amplify or attenuate social norms in dictator decisions. Recent replications in online environments, which inherently enhance anonymity through digital interfaces, confirm these patterns post-2020. For example, virtual dictator games conducted remotely show proportional giving declining with larger virtual endowments, similar to lab findings, while online platforms exhibit similar baseline sharing rates around 20-25% under standard protocols. These results validate the robustness of stakes and anonymity effects across technological contexts, with baseline sharing rates around 20-25% persisting under standard protocols.

Effects of Communication and Resource Source

Communication between the dictator and recipient in the Dictator game significantly influences allocation decisions, primarily by enhancing empathy and social pressure. When recipients are allowed to send a direct plea or "ask" for a share of the endowment, average giving increases substantially compared to a standard no-communication baseline; for instance, giving rises from approximately 15% of the endowment to 24% or more. Subtle interpersonal cues, such as the presence of schematic watching eyes, also modestly boost generosity, with studies showing an increase of around 5-10% in donations under double-blind protocols, particularly among male dictators where giving can double relative to a neutral control. These effects stem from heightened empathy elicited by messages, which prompt dictators to consider the recipient's perspective and reduce selfishness. The source of the endowment similarly shapes behavior, with dictators exhibiting greater selfishness when resources are earned rather than provided as a windfall. In experiments where dictators must complete a real-effort task to obtain their endowment, average giving drops to about 3% compared to 21% in windfall conditions, reflecting stronger feelings of entitlement or property rights over earned funds. Endowments perceived as "dirty money"—such as those originating from an unethical or tainted source—further suppress giving, as exposure to such resources lowers moral standards and promotes unfair allocations in the game. For earned or windfall endowments, mechanisms like moral licensing may play a role, where prior effort justifies retaining more for oneself, contrasting with the empathy-driven responses to communication. Recent research has extended these insights to loss-framed resources, where dictators face the decision of dividing a potential loss rather than a gain. In such setups, giving decreases markedly, with dictators retaining about 10% more of the resource on average (demanding €7.01 out of €10 versus €6.35 in gain frames), indicating heightened selfishness under loss aversion.^[25] This pattern interacts briefly with anonymity, as reduced cues amplify selfish tendencies in loss contexts but are mitigated by direct communication.^[25]

Effects of Environmental and Technological Contexts

Studies conducted on platforms like Amazon Mechanical Turk (MTurk) have demonstrated that baseline sharing behavior in the Dictator game remains similar between online and laboratory settings, with dictators typically allocating around 25% of their endowment on average in post-2020 online experiments. However, online implementations often experience higher dropout rates compared to controlled lab environments, which can introduce selection biases and reduce sample reliability. This elevated attrition is attributed to the lack of supervision and incentives in remote settings, potentially skewing results toward more motivated participants.^[26]^[27] Virtual reality (VR) environments enhance immersion in the Dictator game, influencing prosocial behavior through increased spatial presence and emotional engagement. A 2023 study found that in formal virtual office VR setups, equal splits occurred in approximately 40% of cases compared to 60% in media-lab controls, with dictators retaining about 10% more due to reduced empathy in structured settings. Surreal or non-office VR settings (e.g., cosmos) particularly amplified fairness perceptions, leading to more generous allocations and higher equal splits than formal virtual offices, which mirrored lab-like restraint.^[28] Interactions involving non-human recipients, such as robots or AI agents, generally result in lower giving in the Dictator game. Recent 2025 research indicates that dictators allocate only about 15% to robotic recipients compared to 25% for human ones, reflecting reduced empathy toward artificial entities. This disparity persists even when resources shift from money to alternatives like electricity sharing, where judgments of fairness remain lower for robot-involved scenarios, highlighting boundaries in extending prosocial norms to technology.^[29]^[30] The introduction of potential losses in the Dictator game, such as downside risk for the recipient, significantly reduces sharing due to strategic ignorance, where dictators avoid information that might compel greater generosity. A 2024 study published in Frontiers in Psychology showed that this setup led to lower allocations as participants opted for ignorance to mitigate empathic concerns, underscoring how risk alters the game's pure altruism measure. Such modifications reveal contextual sensitivities beyond standard gain-framed protocols.^[31]

Variants and Extensions

Reverse and Third-Party Variants

The reverse variant of the dictator game, often called the taking game, inverts the standard structure by allowing the recipient to take a portion of the dictator's endowment instead of the dictator deciding how much to give from their own. This modification examines whether generosity in the standard game reflects altruism or is an artifact of the choice set that precludes taking options, thereby testing entitlement norms and potential greed. In experimental implementations, subjects in the taking treatment typically take modest amounts, suggesting that self-interest dominates when taking is feasible but is tempered by social norms against excessive greed.^[32] In the third-party variant, a neutral player serves as the allocator, deciding how to divide an endowment between two other participants without any personal payoff implications, thereby eliminating direct self-interest in the outcome. This setup isolates impartial preferences for fairness and social welfare, as the allocator acts as an observer judging distributions for others. Studies show that third-party allocators exhibit higher generosity compared to standard dictator roles, with choices favoring more equal splits, driven by concerns for minimizing inequality rather than personal gain.^[33] A further extension introduces third-party punishment, where a neutral observer can impose costly sanctions on the allocator after the division is made, at a personal cost to the punisher. This hybrid incorporates the potential for negative reciprocity and tests how the fear of impartial sanctions influences selfish behavior versus norms of entitlement. The punishment option leads to sanctions on unfair allocations, highlighting the role of enforceable fairness standards in curbing pure self-interest.^[34] Collectively, these variants probe the boundaries of selfishness by manipulating roles and consequences, distinguishing intrinsic altruism from context-dependent responses to entitlement and reciprocity norms.

Modern Adaptations

In recent years, the Dictator game has been adapted to benchmark the fairness and social decision-making capabilities of large language models (LLMs). A 2025 study published in Scientific Reports introduced a publicly available benchmark comprising 106 experimental instructions drawn from 38 prior research articles across 12 countries, focusing on LLMs' ability to predict human behavior in Dictator game scenarios. This benchmark evaluates LLMs on both weak tests (comparing predicted average giving rates to human data) and strong tests (matching frequency distributions of giving decisions in 10% increments). For instance, GPT-4 overestimated average human giving at 42.6% compared to the actual 30.7%, highlighting LLMs' tendency to assume greater altruism than observed in human play.^[35] Human-AI hybrid variants have explored allocations where human dictators decide resources for AI or robot recipients, often revealing reduced generosity linked to perceptions of the recipient's non-sentience. A 2025 experiment in Computers in Human Behavior Reports examined third-person judgments on human-robot Dictator games, where participants allocated resources like money, electricity, food, or tools to robot recipients. Results showed significantly lower sharing of human-centric resources (e.g., money or food, with odds ratios as low as 0.12) compared to robot-suited ones like electricity (OR = 0.19), suggesting allocations reflect assumptions about robots' lack of human-like needs or agency.^[29] Field adaptations have translated the Dictator game into real-world charitable contexts, such as online donation platforms, to study prosocial behavior outside lab settings. For example, an online experiment integrated Dictator-like choices into an interface where participants chose between selfish and altruistic allocations to charities like UNICEF, yielding a baseline donation rate of about 42% under anonymity, contrasting lab meta-analyses showing averages of 20-30%. These versions emphasize ecological validity by using actual monetary transfers to nonprofits.^[36]^[16] Virtual reality (VR) implementations provide immersive environments for fairness training, enhancing the game's ecological realism. A study demonstrated VR-mediated Dictator games in simulated office settings, where interactions led to less equal sharing (e.g., dictators offered 25% less on average) than text-based controls, attributed to heightened social presence and nonverbal cues. Such adaptations are increasingly used in training programs to foster empathy and equitable decision-making in professional contexts.^[28] Multi-round digital versions, though uncommon due to the game's one-shot nature, appear in online platforms to examine repeated play without reputation effects. Experimental software like oTree enables anonymous, app-based iterations where endowments vary per round, maintaining independence to isolate intrinsic preferences; findings indicate stable giving across rounds, avoiding learning confounds seen in strategic games. Charity variants direct allocations to nonprofits instead of anonymous individuals, often yielding higher average giving (around 36% as of 2022 meta-analysis) compared to standard anonymous recipients, highlighting the influence of charitable framing on prosociality.^[6]

Applications and Implications

In Behavioral Economics and Psychology

In behavioral economics, the Dictator game serves as a key tool for examining deviations from narrow self-interest, revealing that participants frequently allocate positive amounts to recipients despite no strategic incentive to do so, thus challenging traditional rational choice models. This behavior informs the development of social preference models, such as the inequity aversion framework proposed by Fehr and Schmidt, which posits that individuals dislike not only disadvantageous but also advantageous inequality, leading dictators to share in order to minimize perceived unfairness in outcomes.^[37] The model's predictions align with empirical patterns where dictators avoid extreme selfishness, providing a foundation for understanding how fairness concerns shape economic decisions beyond pure utility maximization.^[16] A seminal meta-analysis by Engel synthesizes data from over 100 experiments, finding that dictators offer an average of approximately 28% of the endowment in standard conditions, underscoring the robustness of these social preferences across contexts.^[16] These insights have practical implications for policy design, where understanding intrinsic motivations for sharing helps craft incentives that promote prosocial outcomes, such as in welfare programs or tax structures that appeal to fairness norms rather than solely financial penalties.^[16] In psychology, the game measures prosociality by isolating unilateral giving decisions, with higher allocations linked to traits like affective empathy, where emotional concern for others predicts greater sharing independent of cognitive perspective-taking.^[38] Studies on moral development, particularly with children, use the Dictator game to track how sharing emerges with age and correlates with empathy; for instance, preschoolers' allocations increase when primed with moral emotions like sympathy, reflecting early sensitivity to others' needs and fairness ideals.^[39] This developmental trajectory highlights the game's utility in linking prosocial behavior to broader psychological processes of moral reasoning. Organizational applications leverage the game to predict workplace generosity, such as in bonus allocations, where individuals in authority roles (analogous to dictators) exhibit similar fairness biases, influenced by hierarchical relationships that amplify or dampen sharing based on perceived status differences.^[30] For example, managers allocating resources to subordinates often mirror lab patterns, giving more when social distance is low, which informs strategies for fostering equitable incentive structures in firms.^[30]

In Emerging Fields

In neuroeconomics, functional magnetic resonance imaging (fMRI) studies have revealed that unfair resource allocations in the Dictator game activate the anterior insula, a brain region associated with processing inequity and negative affect, as observed in experiments from the 2010s where proposers deciding on unequal splits showed heightened insula responses compared to fair offers.^[40] More recent research since 2022 has extended this to empathy circuits, demonstrating that social structure influences altruistic decisions in the game through activations in regions like the temporoparietal junction and ventromedial prefrontal cortex, linking empathy to reduced selfishness in resource sharing.^[41] These findings underscore how neural mechanisms of fairness and social cognition drive behavior beyond purely economic incentives.^[42] Applications in AI ethics have leveraged the Dictator game to benchmark large language models (LLMs) for biases in decision-making, revealing systematic deviations from human norms that inform algorithmic fairness. For instance, 2025 studies show LLMs like GPT-4 often overestimate human altruism in the game, predicting average sharing at 42.6% versus humans' 30.7%, highlighting an optimistic bias that could amplify inequities in AI-driven resource allocation systems.^[35] Such benchmarks emphasize the need for prompt engineering and fine-tuning to align AI with human-like fairness considerations in ethical deployments.^[43] In social robotics, the Dictator game tests human-AI cooperation by examining sharing behaviors toward robotic recipients, with studies indicating substantially lower allocations to robots—particularly humanoid ones—compared to human counterparts due to perceived lack of reciprocity or agency.^[30]^[44] For example, a 2020 study found offers to AI recipients approximately 50 percentage points lower than to humans, informing human-robot interaction (HRI) design and suggesting interventions like anthropomorphic cues to foster equitable collaboration in shared environments. This approach has also been extended to therapeutic contexts, such as studying sharing with robots in children with autism spectrum disorders.^[45] Beyond these areas, the game has been adapted for climate policy simulations, where pro-environmental behaviors correlate with higher sharing in modified versions; for example, climate change distress predicts greater donations in Dictator tasks tied to environmental protection, supporting models of collective action for sustainability.^[46] In gender equity research, 2025 volunteer studies with over 1,100 participants confirm women exhibit significantly higher generosity in the game, transferring about 7.5% more than men after controlling for roles and stakes, aiding analyses of systemic biases in resource distribution.^[47]

Criticisms and Limitations

Methodological Challenges

One prominent methodological challenge in dictator game experiments is the overrepresentation of WEIRD (Western, Educated, Industrialized, Rich, and Democratic) samples, which limits external validity. Studies predominantly draw from university students in high-income countries, particularly the United States, leading to results that may not generalize to broader populations. For instance, cross-cultural analyses reveal substantial variation in dictator offers across 15 small-scale societies, with WEIRD participants exhibiting unusually high prosociality—such as mean offers nearly double those of groups like the Hadza or Tsimane—due to cultural differences in fairness norms and task interpretation.^[48] This bias arises because WEIRD individuals often conceptualize the game through individualistic lenses atypical of global human behavior, potentially skewing inferences about altruism and equity preferences.^[48] Demand effects pose another issue, where participants infer the experimenter's hypothesis and adjust behavior to align with perceived expectations, inflating giving rates. In dictator games, this manifests as heightened generosity when the setup implies prosocial norms, such as in standard "give" formats versus "take" variants. Experimental tests using sequences of give and take games with 215 participants found that while giving decreases under anonymity or take options, a residual 45% of dictators transferred above the minimum, linked more to preferences for procedural fairness (e.g., compensating for the recipient's lack of choice) than pure demand compliance.^[49] Mitigation strategies include deception controls, like varying instructions or double-blind procedures, to reduce hypothesis guessing without fully eliminating the effect.^[49] The shift to online platforms post-2020 has enhanced scalability by increasing participant access and reducing costs, but it introduces challenges like reduced experimental control and data contamination from bots. Easier recruitment via social media and crowdsourcing sites has led to a surge in fraudulent responses, with bots generating plausible but inconsistent data—such as rapid submissions or AI-crafted open-text answers—that compromise validity. In behavioral research, including economic games, this has necessitated rigorous filtering (e.g., attention checks and IP verification), yet up to 75% of responses in some surveys remain invalid, undermining the reliability of large-scale dictator game replications.^[50] Ethical concerns arise from using real stakes, which can induce financial stress, particularly among low-income or vulnerable participants, and require careful handling of informed consent under anonymity protocols. Monetary incentives, standard in dictator games to motivate realistic decisions, risk undue influence by pressuring participation despite potential discomfort from unequal outcomes or loss aversion. Guidelines emphasize ensuring voluntary consent without coercion, protecting participant welfare from stress (e.g., via debriefing), and maintaining anonymity to prevent social repercussions, though high stakes may still exacerbate anxiety in resource-scarce groups.^[51]

Theoretical and Interpretive Debates

One central debate in interpreting Dictator game results concerns the distinction between pure altruism and reciprocity-driven motives, even in ostensibly anonymous settings. While the game is designed to isolate unconditional generosity by removing recipient influence, critics argue that observed giving often reflects residual expectations of reciprocity or social norms rather than intrinsic altruism. For instance, experiments demonstrate that dictator offers can reverse dramatically when stakes are altered or anonymity is imperfectly enforced, suggesting that generosity may be an experimental artifact influenced by impure motives like anticipated future interactions or reputational concerns, rather than selfless giving.^[32] This confusion challenges the game's validity as a pure measure of altruism, prompting calls for disentangling these motives through variants that explicitly control for reciprocity cues.^[52] Cultural interpretations of Dictator game outcomes have also sparked controversy, particularly regarding the universality of fairness norms. Joseph Henrich and colleagues have critiqued the overreliance on Western, Educated, Industrialized, Rich, and Democratic (WEIRD) samples, arguing that they lead to erroneous assumptions of universal prosociality, as evidenced by cross-cultural studies showing wide variation in giving levels tied to local institutions and evolutionary histories. Post-2020 analyses have pushed back against overgeneralizations from such data, highlighting how cultural determinism—where societal norms rigidly dictate behavior—overlooks individual agency and contextual fluidity in non-WEIRD populations.^[48] These debates underscore the risk of ethnocentric bias in extrapolating lab findings to global human behavior. Recent gender findings in the Dictator game have reignited discussions on whether women's consistently higher giving rates stem from biological predispositions or socialization processes. A 2025 high-powered laboratory study with 1,161 participants found that women transfer approximately 7.5% more of the endowment than men, even after controlling for risk preferences, reasoning ability, personality traits, and other factors, attributing this to potential gender-specific prosocial tendencies mediated by emotions and traits like agreeableness.^[47] However, subsequent 2025 research debates the causes, with some evidence suggesting socialization effects—such as norms encouraging female empathy—amplify giving, while others invoke biological markers like prenatal hormone exposure to explain persistent differences across contexts.^[53]^[54] This interpretive tension highlights the need for integrated models combining evolutionary biology and cultural learning. Finally, the game's implications for policy applications remain contentious due to its limited external validity. Overreliance on lab results for designing incentives like charitable matching programs risks ignoring real-world contexts, as dictator giving often fails to predict actual donations; for example, experiments show that while lab participants give around 20-30% of endowments, field behaviors drop significantly without immediate social pressure or personal stakes.^[55] This disconnect urges caution in translating game insights to policy, emphasizing hybrid approaches that incorporate ecological validity.

References

[1]
Economic Games: An Introduction and Guide for Research | Collabra
Feb 26, 2021 · In this game, a dictator can freely decide how to distribute an endowment between him/herself and a recipient, who cannot react to the ...The Potential of Economic... · A Brief Introduction into... · II. Economic Games: An...
[2]
Dictator games: a meta study | Experimental Economics
Mar 14, 2025 · Over the last 25 years, more than a hundred dictator game experiments have been published. This meta study summarises the evidence.
[3]
On the Interpretation of Giving in Dictator Games
The dictator game represents a workhorse within experimental economics, frequently used to test theory and to provide insights into the prevalence of social ...Missing: definition | Show results with:definition
[4]
Fairness in Simple Bargaining Experiments - ScienceDirect.com
The outcomes of the ultimatum and dictator games with pay are significantly different, implying that fairness, by itself, cannot explain the observed behavior.
[5]
Dictator Games: A Meta Study by Christoph Engel :: SSRN
Mar 13, 2010 · Abstract. Over the last 25 years, more than a hundred dictator game experiments have been published. This meta study summarizes the evidence.
[6]
Bridging the gap between the economics lab and the field: Dictator ...
May 29, 2023 · We report three experiments where we present the same participants with different versions of the dictator game and with two different field situations.Missing: disciplinary | Show results with:disciplinary
[7]
https://www.mdpi.com/journal/games/special_issues/Dictator_Games
[8]
[PDF] 6 Extensive Games with Perfect Information: Illustrations
Several related games have been studied, among them the dictator game and the impunity game (Exercise 183.3). In the subgame per- fect equilibria of these games ...
[9]
https - DOI
No information is available for this page. · Learn why
[10]
[PDF] Dictator Games: A Meta Study - Max-Planck-Gesellschaft
In dictator games, the dependent variable is (quasi)continuous. Dictators may give any frac- tion between 0 and 100 % of their endowments, albeit frequently ...
[11]
[PDF] game instructions - Department of Psychology | UZH
GAME INSTRUCTIONS – DICTATOR GAME. 4. DICTATOR GAME. ALLOCATOR. This task is about dividing money between yourself and another person to whom you are randomly.
[12]
[PDF] Dictator games: A meta study - EconStor
Kahneman invented the game as part of his programme that turned textbook assumptions into behavioural hypotheses. While normally a sizeable fraction of ...
[13]
https://www.econstor.eu/bitstream/10419/38870/1/622756745.pdf
[14]
Dictator games: a meta study | Experimental Economics
May 20, 2011 · ... dictator game data. It shows how Tobit models (assuming that dictators would even want to take money) and hurdle models (assuming that the ...
[15]
[PDF] Behavioral experiments in 15 small-scale societies
To address this, we undertook a cross-cultural study of behavior in ultimatum, public goods, and dictator games in a range of small-scale societies exhibiting a ...
[16]
Individualism, Collectivism, and Allocation Behavior: Evidence from ...
Feb 6, 2023 · In the dictator game, we found that participants in the collectivism-priming condition reported a slightly higher mean offer than in the ...
[17]
https://www.journals.uchicago.edu/doi/10.1086/678181
[18]
Children's Sharing Behavior in Mini‐Dictator Games: The Role of In ...
Nov 7, 2016 · Each child played two mini-dictator games against an in-group member (friend) and an out-group member (stranger) to divide four stickers.Missing: peaks | Show results with:peaks
[19]
Measuring Altruistic Behavior in Surveys: The All-or-Nothing Dictator ...
Aug 6, 2025 · Bekkers (2007) found that more educated individuals tend to give more in dictator games, likely because those with higher education levels have ...
[20]
[PDF] Chapter 18 - Russell Sage Foundation
... demographic differences that we see in figures 18.1 to 18.6 appear to have no impact on prosocial behavior in the dictator game. Although these DG results ...
[21]
https://www.researchgate.net/publication/5022883_Measuring_Altruistic_Behavior_in_Surveys_The_All-or-Nothing_Dictator_Game
[22]
https://www.russellsage.org/sites/default/files/ensminger/13521-18_CH18_4thPgs.pdf
[23]
https://doi.org/10.1016/j.beproc.2015.04.003
[24]
Covid-19 and altruism: a meta-analysis of dictator games
Aug 9, 2025 · We find that dictators on average share approximately 22% of their endowment. We also find robust evidence that earned endowment reduces ...
[25]
Comparing online and lab methods in a problem-solving experiment
Disadvantages included high dropout rate for online participants; however, combining the online experiment with the department subject pool worked well.Missing: behavioral economics
[26]
Lab vs online experiments: No differences - ScienceDirect.com
Results suggest that there are no differences between conditions, except for a larger share of online subjects donating nothing in the Dictator Game. •. Overall ...
[27]
Setting the stage: Fairness behavior in virtual reality dictator games
As expected, the majority of the dictators in our one-shot dictator game chose the equal divide. The experiment was conducted at a small university and the ...
[28]
Third-person perspectives on human-robot dictator game outcomes
Research indicates that people generally express a high willingness to help robots, with positive attitudes toward robots being a contributing factor (Potinteu ...<|control11|><|separator|>
[29]
We, Them, and It: Dictator Game Offers Depend on Hierarchical ...
Nov 22, 2020 · In summary, the aim of this study is to examine the influence of different hierarchical levels on the distribution of money in the dictator game ...
[30]
Seemingly altruistic behavior and strategic ignorance in a dictator ...
The purpose of the present study is to use a dictator game with loss potential that facilitates player empathic concern to investigate the emergence of ( ...<|control11|><|separator|>
[31]
Dictator game giving: altruism or artefact? | Experimental Economics
Sep 29, 2007 · Experimental dictator games have been used to explore unselfish behaviour. Evidence is presented here, however, that subjects' generosity can be reversed.
[32]
[PDF] Third-party punishment and social norms
We denote this experiment as third party punishment in the dictator game (TP-. DG). Player A had an endowment of 100 points and could transfer 0, 10, 20, 30, 40 ...Missing: Froese 2010
[33]
A publicly available benchmark for assessing large language ...
Jul 1, 2025 · Here we introduce a novel, publicly available, benchmark to test LLM's ability to predict how humans balance monetary self-interest and the interest of others.
[34]
Moral suasion and charitable giving | Scientific Reports - Nature
Dec 1, 2022 · UNICEF receives $0 if the dictator chooses A, and $3 if they choose option B. Before making their decision, dictators receive (from another ...Missing: VR training
[35]
Dictator Game with different amounts randomly generated every round
I want to play 10 iterations of Dictator Game where a random amount of money will be distributed among the Dictator and the algorithm.
[36]
https://www.nature.com/articles/s41598-022-24944-6
[37]
Explaining altruistic sharing in the dictator game - ScienceDirect.com
Our study focuses on empathy as a predictor for altruistic sharing in the dictator game and thereby complements economic explanations. Although the focus of the ...
[38]
Preschoolers' allocations in the dictator game: The role of moral ...
This study explores how 3–5-year-old children allocate resources in the dictator game, and whether participants' understanding of moral emotions predicted ...
[39]
Brain responses to self- and other- unfairness under resource ...
Aug 15, 2024 · Common activation of unfairness to self and to others was found in the dorsal anterior insula and anterior cingulate cortex, with stronger ...
[40]
Neural signatures underlying the effect of social structure on ...
Jul 15, 2025 · First study to identify neural mechanisms by which social structure influences empathy and altruism, beyond direct relationships.
[41]
Neural mechanisms of costly helping in the general population and ...
May 21, 2024 · Encouraging evidence stems from Ioumpa et al. who found that mirror-sensory synesthetes donate more money to a stranger in a dictator game, but ...
[42]
https://www.nature.com/articles/s41598-024-62422-3
[43]
Behavioral-economic Games with Commercially Available Robots
Mar 11, 2024 · Using commercially available robots, we showed participants photographs of 18 robots, and had them play a dictator game against these robots ...
[44]
Panic internally, act sustainably: Climate change distress predicts ...
CCD predicted PEB in the work for environmental protection task and a dictator game. •. Younger individuals and women (vs. men) experienced the highest levels ...
[45]
Gender differences in dictator giving: A high-power laboratory test
We find that women behave in a significantly more generous way than men: after controlling for the factors mentioned above, females transfer 7.5 percentage ...
[46]
[PDF] The weirdest people in the world? - Description
with other colleagues: Henrich et al. 2004; 2005) about the cross-cultural variation in the Ultimatum, Dictator, and Public. Goods games can be at least ...
[47]
[PDF] Experimenter demand effect or preference over the rules of the game?
To date, several studies find that dictator game giving breaks down under conditions designed to increase dictators' anonymity or if an option to take money ...Missing: pleas | Show results with:pleas
[48]
It's raining bots: how easier access to internet surveys has created ...
Jun 1, 2025 · Bots operate based on algorithms that allow them to navigate, interact and execute actions on digital platforms with or without human ...
[49]
[PDF] Chapter 9 - Ethics and Experimental Economics - Barnard College
Respect for persons asks that the autonomy of indi- viduals be respected, that they give informed consent to participate in the experiment, and that individuals ...
[50]
[PDF] Ethical Concerns of Paying Cash to Vulnerable Participants
Dec 21, 2020 · Abstract. The aim of the paper is to discuss the ethical issues related to financial payments. The article compares.<|separator|>
[51]
Fairness, Reciprocity, and Stakes in Variants of the Dictator Game
In two experiments, the existence and extent of altruistic reciprocity is explored in the context experimental game, "the sequential dictator." Findings show ...