(Investment Game) – Joyce Berg, John Dickhaut, and Kevin McCabe The Public Goods

The architecture of modern economic exchange rests upon a fundamental behavioral paradox. Canonical neoclassical theory, grounded in the axiomatic formalisms of expected utility and game-theoretic subgame perfection, presupposes that self-interested agents interact solely through market mechanisms reinforced by costless contract enforcement or repeated-interaction reputational safeguards. When bilateral exchange occurs in an isolated, one-shot, anonymous setting devoid of legal recourse, standard microeconomic theory delivers an unambiguous prediction: rational opportunism will prevail, commitments will dissolve into pervasive default, and the gains from prospective cooperation will be wholly forfeited. Yet, across civil societies, decentralized economies, and institutional ecosystems, human actors consistently engage in uncollateralized investments, extending credit, committing proprietary resources, and contributing to collective endowments where purely selfish optimization would dictate immediate free-riding. Reconciling this stark divergence between formal deductive models of Homo economicus and the observed realities of social cooperation represents one of the foundational undertakings of modern social science.

In their seminal 1995 publication, “Trust, Reciprocity, and Social History”, published in Games and Economic Behavior, Joyce Berg, John Dickhaut, and Kevin McCabe engineered a methodological and conceptual paradigm shift. By designing an austere, two-stage sequential laboratory experiment—subsequently immortalized as the “Investment Game” or “Trust Game”—the authors isolated the primary microfoundations of non-contractual economic exchange: trust and reciprocity. In doing so, they provided an empirical refutation of the zero-trust subgame perfect Nash equilibrium, demonstrating that trust is an endogenous behavioral asset capable of generating substantial social surplus. Simultaneously, their findings established a direct structural and conceptual bridge to the literature on public goods provision, illuminating how bilateral reciprocal inclinations serve as the decentralized engine underpinning larger-scale collective action dilemmas.

This extensive treatise presents a comprehensive, multifaceted examination of the Investment Game devised by Berg, Dickhaut, and McCabe. Across twelve exhaustive thematic sections, we delineate the historical and theoretical genesis of the paradigm; formalize its mathematical and game-theoretic machinery; dissect its psychological, neurobiological, and evolutionary drivers; and trace its foundational connections to public goods games and voluntary contribution mechanisms. Furthermore, this study evaluates the methodological innovations and critiques that have shaped thirty years of subsequent experimental research, exploring the deep institutional implications of reciprocal trust for contract design, digital peer-to-peer architectures, and macroeconomic performance.

1. Foundations and Historical Context of the Investment Game

1.1 The 1995 Seminal Experiment by Berg, Dickhaut, and McCabe

The publication of “Trust, Reciprocity, and Social History” in 1995 marked a watershed moment in behavioral and experimental economics. Prior to the design formulated by Joyce Berg, John Dickhaut, and Kevin McCabe at the University of Minnesota and the University of Arizona, empirical inquiries into human cooperation had largely been relegated to observational sociology or blended within complex market simulations where strategic reputation, repeat-play dynamics, and informal socialization confounded clean identification. Economists lacked an instrument capable of cleanly separating genuine, non-strategic trust from the forward-looking calculations of repeated games modeled by the folk theorem. Berg, Dickhaut, and McCabe recognized that establishing trust as a primitive behavioral mechanism required an experimental environment stripped of all institutional scaffoldings, repeated-interaction incentives, third-party enforcement, and reputational spillovers.

Their operational methodology established a simple yet elegant sequential protocol. In a strictly anonymous, one-shot laboratory environment, an initial subject (the investor or trustor) received a monetary endowment and was granted the liberty to transmit any portion of it—or none at all—to an anonymous counterpart (the trustee). Crucially, any transferred funds were multiplied by a known constant (specifically tripled) by the experimenters before reaching the trustee, thereby creating a dramatic expansion of the aggregate social surplus. The trustee was then faced with a completely unconstrained choice: retain the entirety of the multiplied funds, or return any fraction back to the investor. By implementing this protocol under rigorous laboratory controls, Berg, Dickhaut, and McCabe transitioned the study of trust from speculative philosophy into an empirically tractable, falsifiable discipline.

The motivation driving this experimental architecture was to isolate whether trust and reciprocity could exist purely as primitive social preferences. By ensuring that interactions were completely non-repeated and conducted between completely decoupled, anonymous participants, the authors closed every conventional theoretical escape hatch that standard non-cooperative game theory utilized to explain cooperation. The results yielded by this structural environment served as an empirical cornerstone that directly challenged prevailing axiomatic models of choice, demonstrating that cooperative surplus generation is an innate feature of human economic agency even in the total absence of exogenous enforcement.

1.2 Behavioral Anomalies and the Critique of Homo Economicus

For the greater part of the twentieth century, neoclassical economics operated upon the foundational paradigm of Homo economicus—an idealized agent characterized by infinite cognitive processing capability, stable and well-ordered preferences, and an unyielding, exclusive drive toward personal material wealth maximization. Under this canonical doctrine, human interactions in strategic settings are governed strictly by self-interest. Game theory, formalizing this doctrine through the concepts of Nash equilibrium and subgame perfection via backward induction, operates on the presumption that any rational agent will exploit any sequential advantage that yields an incremental material payoff, provided no subsequent penalties can be levied against them.

The behavioral anomalies unveiled by Berg, Dickhaut, and McCabe dealt a direct blow to this axiomatic framework. When evaluated through the analytical lens of backward induction, the canonical model yields a stark and definitive prediction: the second mover, possessing no material incentive to part with wealth already in their possession, will return zero dollars to the investor regardless of the magnitude of the initial transfer. Anticipating this total defection, the rational first mover, evaluating the terminal node of the decision tree, will transfer exactly zero dollars. The theoretical equilibrium is a state of universal paralysis: zero investment, zero returns, and complete failure to capture the threefold efficiency gains made available by the experimental production technology.

The laboratory outcomes demonstrated precisely the opposite pattern. Rather than converging toward the zero-trust equilibrium, subjects exhibited high levels of transfer, with investors routinely committing substantial proportions of their endowments, and trustees reciprocating by returning sums that frequently exceeded the baseline principal. This profound empirical contradiction catalyzed the rise of modern behavioral economics. It established that human subjects systematically deviate from pure wealth maximization, displaying robust other-regarding preferences, sensitivity to social norms, and deep-seated aversions to opportunistic exploitation. The Investment Game thus served as an undeniable demonstration that the standard neoclassical behavioral postulate was descriptively deficient when applied to asymmetric, sequential social interactions.

1.3 Divergence from Classical Public Goods and Dilemma Paradigms

Prior to the introduction of the Investment Game, the predominant experimental tools for analyzing human cooperation were simultaneous-move games, notably the Prisoner’s Dilemma and the linear Voluntary Contribution Mechanism (VCM) public goods game popularized by Isaac, Walker, and Thomas. While these paradigms successfully exposed the tensions between individual incentives and collective efficiency, their structural characteristics masked the distinct behavioral mechanisms governing bilateral human relationships. In a standard linear public goods game, $N$ players simultaneously allocate tokens between a private account and a common pool, with contributions to the common pool being multiplied by an efficiency factor and redistributed equally among all group members regardless of their individual input.

The structural divergence between the linear public goods paradigm and the Berg, Dickhaut, and McCabe Investment Game is profound along several dimensions:

• Temporal Asymmetry: The public goods game is characterized by simultaneous choice, where participants make allocation decisions under strategic uncertainty regarding peer behavior. The Investment Game is explicitly sequential, decoupling the choice of vulnerability (the trustor’s transfer) from the choice of fulfillment or exploitation (the trustee’s return).
• Agency and Target Specificity: In a public goods dilemma, an individual’s contribution generates a diffuse, non-excludable positive externality distributed symmetrically across the entire group. In contrast, the Investment Game establishes a targeted, bilateral channel of exposure. The trustor places their welfare exclusively in the hands of a single, identifiable, albeit anonymous, other agent.
• Structural Decoupling of Norms: The linear public goods game conflates cooperation, altruism, and confusion within a single mathematical contribution. Conversely, the Investment Game cleanly disentangles the initial act of trust (incurring vulnerability to create a surplus) from the subsequent act of reciprocity (voluntarily bearing a cost to reward prior benevolence).

By enforcing this asymmetric, sequential architecture, the Investment Game moved beyond generalized collective action to isolate the relational microfoundations of economic value creation. It demonstrated that before large-scale public goods can be sustained through multilateral norms, bilateral interactions must be supported by reciprocal dynamics capable of counteracting the threat of opportunism.

2. Mathematical Formalization and Game-Theoretic Mechanics

2.1 Structural Architecture of the Two-Stage Sequential Game

The formal game-theoretic structure of the Berg, Dickhaut, and McCabe Investment Game can be parameterized as an extensive-form game of complete but imperfect information, played between two risk-neutral or risk-averse agents: Player 1 (the Trustor or Investor) and Player 2 (the Trustee or Responding Agent). Let the interaction proceed through discrete chronological stages over a finite game tree:

At Stage 1, Player 1 is granted an exogenous monetary endowment $M in \mathbb{R}_{++}$, while Player 2 receives an identical baseline endowment $M$, designed to eliminate absolute wealth asymmetry at the outset. Player 1 is endowed with the action set $A_1 = [0, M]$, selecting an investment transfer $X in A_1$. The transfer represents the quantity of capital that Player 1 chooses to expose to the hazard of Player 2’s discretion. The residual $(M – X)$ is retained immediately in Player 1’s private account.

Between Stage 1 and Stage 2, the transfer $X$ traverses an experimental production function, operationalized as an exogenous multiplier factor $k > 1$. In the canonical Berg, Dickhaut, and McCabe implementation, $k = 3$. Player 2 receives the augmented transfer $kX$. At Stage 2, Player 2 observes the exact realized value of $X$ and possesses an action set defined dynamically by the realized transfer: $A_2(X) = [0, kX]$. Player 2 selects a return transfer $Y in A_2(X)$, representing the volume of the multiplied surplus to be remitted back to Player 1.

The terminal payoffs for the two agents, denoted $\Pi_1(X, Y)$ and $\Pi_2(X, Y)$, are defined as follows:

$$\Pi_1(X, Y) = M – X + Y$$

$$\Pi_2(X, Y) = M + kX – Y$$

This structural architecture yields a variable-sum game where the aggregate wealth of the dyad, $W(X, Y) = \Pi_1 + \Pi_2 = 2M + (k – 1)X$, is strictly increasing in $X$. Maximum joint efficiency is achieved if and only if Player 1 chooses $X = M$, expanding total system wealth to $(k + 1)M$.

2.2 Subgame Perfection, Backward Induction, and the Zero-Trust Equilibrium

To identify the standard non-cooperative game-theoretic solution, the game is analyzed using the methodology of backward induction, which identifies the subgame perfect Nash equilibrium (SPNE) in finite extensive-form games. Assume that both players are purely self-interested, evaluating outcomes exclusively according to monotonically increasing material payoff functions: $U_i(\Pi_i) = \Pi_i$.

Consider the subgame initiated at Stage 2 following an arbitrary transfer $X > 0$ by Player 1. Player 2 faces the optimization problem:

$$\max_{Y in [0, kX]} U_2(M + kX – Y) = \max_{Y in [0, kX]} (M + kX – Y)$$

Because the objective function is strictly decreasing in $Y$ across the entire domain $[0, kX]$ for any $X$, the derivative of Player 2’s utility with respect to $Y$ is strictly negative:

$$\frac{\partial U_2}{\partial Y} = -1 < 0$$

Consequently, the unique dominant strategy for Player 2 at this decision node, irrespective of the magnitude of $X$, is the boundary solution:

$$Y^*(X) = 0 \quad \forall X in [0, M]$$

Moving backward to Stage 1, Player 1 possesses perfect knowledge of Player 2’s rationality and utility function, and therefore correctly forecasts Player 2’s best-response function $Y^*(X) = 0$. Player 1’s strategic optimization problem at the initial node degenerates to:

$$\max_{X in [0, M]} U_1(M – X + Y^*(X)) = \max_{X in [0, M]} (M – X + 0) = \max_{X in [0, M]} (M – X)$$

Taking the first-order derivative with respect to $X$ yields:

$$\frac{\partial U_1}{\partial X} = -1 < 0$$

Player 1’s utility is strictly decreasing in $X$. Therefore, the unique subgame perfect Nash equilibrium strategy profile is:

$$(X^*, Y^*(X)) = (0, 0)$$

The resulting equilibrium payoffs are uniquely determined at $(\Pi_1^*, \Pi_2^*) = (M, M)$. This outcome demonstrates an acute form of market failure: the subgame perfect equilibrium is severely Pareto-inefficient. Mutual distrust locks the dyad into a behavioral trap, fully squandering the potential surplus of $(k – 1)M$.

2.3 Surplus Generation and Social Efficiency Metrics

The core tension within the Investment Game centers on the divergence between individual rationality and systemic social efficiency. Social efficiency in this setting can be formalized through an efficiency metric $E in [0, 1]$, defined as the ratio of realized social surplus to the maximum theoretically attainable social surplus:

$$E(X) = \frac{W(X, Y) – W_{\min}}{W_{\max} – W_{\min}} = \frac{(2M + (k-1)X) – 2M}{(2M + (k-1)M) – 2M} = \frac{(k-1)X}{(k-1)M} = \frac{X}{M}$$

Remarkably, the realized social efficiency of the interaction is entirely a linear function of Player 1’s transfer $X$, completely independent of Player 2’s return choice $Y$. The trustee’s return decision $Y$ operates purely as a zero-sum, non-distortionary distributional transfer, determining the ex-post equity of the surplus allocation without altering its aggregate volume. If $X = 0$, $E = 0$, representing the complete collapse of collective value creation. If $X = M$, $E = 1$, realizing maximal social efficiency.

This mathematical equivalence links the Investment Game’s multiplication factor $k$ directly to the production functions governing public goods environments. In standard linear public goods games, the marginal per capita return (MPCR), denoted $\alpha$, dictates the individual return on contributions to the collective pool relative to the private opportunity cost. For an $N$-player public goods game, social efficiency requires $N\alpha > 1$, while individual opportunism is incentivized whenever $alpha < 1$. In the two-player Investment Game, the multiplier$k$ plays an analogous role: since $k = 3 > 1$, social welfare mandates full investment, yet because Player 1 captures none of the generated surplus unless Player 2 reciprocates ($Y/X > 1$), the individual marginal return on investment under pure self-interest is zero. The game formalizes an essential economic reality: the realization of productive efficiency requires one agent to assume private exposure for the generation of a common surplus.

3. Conceptualizing Trust and Reciprocity in Economic Theory

3.1 Berg, Dickhaut, and McCabe’s Operational Definitions

A seminal contribution of the 1995 paper was the rigorous operationalization of concepts that had long resisted precise analytical treatment within economics: trust and reciprocity. Prior to this work, trust had frequently been treated as a vague cultural residue, an ephemeral moral sentiment, or an indistinguishable facet of generalized risk tolerance. Berg, Dickhaut, and McCabe anchored these phenomena directly within verifiable choice architectures.

The authors defined trust as the voluntary commitment of resources to an exchange partner in an environment where such commitment exposes the decision-maker to the risk of opportunistic exploitation, with no binding legal recourse, under the expectation that the counterpart will act benevolently rather than exploitatively. Operationally, trust is embodied in the decision by Player 1 to select $X > 0$. This definition demands the presence of two indispensable structural components:

• The vulnerability requirement: The transfer must expose the trustor to potential net loss (if $Y < X$, the investor is worse off than if no transfer had occurred).
• The surplus-generating expectation: The act is taken under the plausible anticipation that mutual gain will be achieved through the counterpart’s voluntary cooperation.

Conversely, reciprocity was operationalized as the contingent, costly response of a secondary agent to a prior beneficial action initiated by another. It is fundamentally distinct from unconditional benevolence. Player 2’s return transfer $Y > 0$ constitutes reciprocity if and only if it is causally conditioned upon the receipt of $X > 0$, representing an unforced willingness to surrender material resources to reward the trustor for their initial vulnerability. Berg, Dickhaut, and McCabe explicitly delineated these definitions from strategic reputation-building, emphasizing that in a one-shot, double-blind environment, reciprocity must stem from internalized behavioral norms rather than forward-looking self-interest.

3.2 Distinguishing Pure Altruism from Strategic Reciprocity

To establish the empirical autonomy of trust and reciprocity, it was necessary to delineate them cleanly from pure altruism. In economic terminology, pure altruism or unconditional warm-glow giving (as modeled by James Andreoni) reflects preferences where an agent derives utility directly from the absolute consumption or payoff level of another player, independent of prior interactions, history, or intent:

$$U_i = u_i(\Pi_i) + \gamma_i u_j(\Pi_j), \quad \gamma_i > 0$$

Under pure altruism, a decision-maker transfers wealth simply because they care about the recipient’s well-being. This behavior is captured experimentally in the Dictator Game (Kahneman, Knetsch, and Thaler, 1986; Forsythe et al., 1994), where an agent unilaterally divides an endowment with a passive recipient who possesses no capacity to respond.

Strategic reciprocity operates upon fundamentally different psychological and mathematical premises. Reciprocal preferences are intrinsically state-dependent and action-contingent. An individual displaying positive reciprocity experiences utility from increasing the payoff of another agent if and only if that agent has previously demonstrated benevolent intentions or generated a material benefit for them. Under reciprocity, Player 2’s utility function incorporates an explicit relational weighting parameter $\theta(X)$ that is endogenous to Player 1’s prior transfer:

$$U_2 = \Pi_2 + \theta(X) \cdot \Pi_1$$

Where $\theta(X) \leq 0$ when $X = 0$, and $\theta(X) > 0$ when $X$ reflects a substantial act of trust. Reciprocal transfers cannot be reduced to unconditional altruism, because the trustee’s willingness to transfer wealth $Y$ collapses to zero in the absence of an intentional, beneficial transfer $X$ from the trustor. Econometric evaluations that compare Dictator Game distributions with Trustee return distributions systematically confirm that while baseline altruism accounts for a minor fraction of transfers, the vast majority of returned capital in the Investment Game is explained exclusively by positive reciprocity.

3.3 Risk Preferences versus Social Preferences in Investment Decisions

A crucial theoretical dispute emerging from the Investment Game concerns whether the trustor’s decision $X$ reflects social preferences or merely standard risk tolerance under subjective probability distributions. If Player 1 views the trustee simply as a stochastic mechanism that yields returns with some subjective probability distribution $P(Y | X)$, the decision to invest could be characterized as an ordinary lottery choice under uncertainty, fully explicable via standard Von Neumann-Morgenstern expected utility theory without invoking social norms.

Decades of experimental inquiry have thoroughly dismantled this reductionist interpretation. Controlled experiments comparing decisions in the Investment Game with functionally identical “Risky Dictator Games”—where Player 2 is replaced by a computerized randomizing device calibrated to match the empirical return distribution of human trustees—reveal persistent and statistically profound discrepancies. Subjects consistently transfer significantly less wealth to human trustees than they commit to mechanical lotteries exhibiting identical payoff odds. This behavioral gap demonstrates the phenomenon of betrayal aversion: human decision-makers experience an acute psychological disutility from being actively exploited by a peer that vastly exceeds the disutility associated with suffering an identical financial loss through impersonal, state-of-nature bad luck.

Consequently, the decision to trust cannot be conflated with mere financial risk tolerance. Trust represents a distinct social preference involving the simultaneous evaluation of interpersonal risk, subjective beliefs regarding counterpart moral obligations, and vulnerability to betrayal. In asymmetric information environments, this distinction becomes paramount, as the perceived intentionality of the counterpart profoundly alters the investor’s baseline willingness to commit resources to collective enterprise.

4. Experimental Design and Protocol Innovations

4.1 Double-Blind Protocols and Anonymity Safeguards

Recognizing the radical nature of their empirical claims, Berg, Dickhaut, and McCabe recognized that any observed deviations from subgame perfection could easily be dismissed by orthodox theorists as methodological artifacts resulting from experimenter demand effects or implicit reputational concerns. If subjects suspected that the experimenter could observe their individual allocations, their choices might reflect a desire for social approval or fear of moral judgment rather than authentic other-regarding preferences. Similarly, if subjects believed their counterparts might identify them following the session, fear of real-world retaliation or desire for post-experimental socialization could generate spurious cooperation.

To eliminate these pervasive confounds, the authors implemented an uncompromising double-blind experimental protocol, building upon the methodological frontiers pioneered by Elizabeth Hoffman, Kevin McCabe, and Vernon Smith. The architectural safeguards ensured that:

• No participant could ever identify the specific individual with whom they were paired;
• The experimenters themselves could never link an individual subject’s identity, physical person, or demographic profile to their specific decision-making record or cash payouts.

The physical logistics of this double-blind design were intricate. Subjects were recruited in cohorts, physically isolated into two distinct rooms (Room A for Investors, Room B for Trustees), and identified solely by randomly assigned numeric keys. Transfers were enacted via physical, opaque envelopes containing standardized cash denominations and blank control sheets. When an investor allocated a transfer, the cash was deposited into an inner envelope, sealed, and passed through a physical blind drop box monitored by a disinterested third-party proctor who had no access to master key rosters. The experimenters received the envelopes anonymously, multiplied the enclosed currency, and delivered them to Room B trustees under identical blind procedures. Payouts at the conclusion of the experiment were distributed via private, keyed lockers or opaque individual envelopes retrieved out of sight of both peers and researchers. By demonstrating that substantial trust and reciprocity persisted under these extreme, sterile conditions of total anonymity, Berg, Dickhaut, and McCabe proved conclusively that cooperative behavior in economic games is not an illusion generated by experimental scrutiny.

4.2 Social History Treatment: Information Conditions

A second foundational innovation introduced in the 1995 study was the explicit experimental manipulation of social information, known as the “Social History” treatment. The authors recognized that human economic culture does not emerge in an institutional vacuum; real-world economic actors operate with access to historical precedents, cultural norms, and aggregated feedback regarding the behavioral tendencies of their peers. To analyze whether awareness of societal behavioral patterns would stabilize or erode trust, the authors instituted a comparative institutional framework.

In the Social History treatment, a newly recruited cohort of completely naive subjects was provided with the full, unvarnished empirical distribution of choices made by the participants in the previous baseline sessions. Before making their allocations, both investors and trustees in the Social History condition were presented with a summary report detailing the exact amounts transferred by previous Room A participants and the precise amounts returned by Room B counterparts. Standard non-cooperative game theory predicts that social history should have zero impact on behavior: since the backward induction logic for the subgame perfect Nash equilibrium is invariant to historical observation, rational agents possessing common knowledge of rationality should continue to play $X = 0$ and $Y = 0$.

Conversely, behavioral theories hypothesized that social history could either serve as an informational catalyst for normative erosion—if subjects observed evidence of positive free-riding by trustees—or operate as a focal coordination mechanism that legitimizes and stabilizes reciprocal exchange. The empirical implementation of this condition represented one of the earliest laboratory demonstrations of how macro-level social data interfaces with micro-level strategic choice, illuminating the evolutionary transmission of economic norms across distinct generational cohorts.

4.3 Payoff Structures and Empirical Implementation

The empirical parameters selected by Berg, Dickhaut, and McCabe were carefully calibrated to ensure genuine economic salience. The baseline experimental endowment was set at $M =$10$ per subject, administered alongside a nominal participation fee. In 1995 currency, a ten-dollar endowment represented a non-trivial sum for undergraduate subjects, roughly equivalent to several hours of campus labor, ensuring that choices carried real, tangible opportunity costs and satisfying the rigorous requirements of induced value theory.

The multiplication factor was set definitively at $k = 3$. This choice yielded an exceptionally steep efficiency frontier. If an investor transferred their full $10 endowment, the trustee received$30, which, combined with the trustee’s own initial $10 baseline endowment, yielded a total allocatable pool of$40 for Player 2. The physical methodology utilized ten one-dollar bills inside the transfer envelopes, forcing transfers to occur in discrete integer increments:

$$X in {0, 1, 2, dots, 10}$$

Trustees were similarly constrained to integer returns:

$$Y in {0, 1, 2, dots, 3X}$$

The instructions provided to participants were meticulously neutral, deliberately purged of moralizing rhetoric, emotional cues, or cooperative terminology. The terms “trust,” “reciprocity,” “investment,” “partner,” or “exploitation” were scrupulously avoided. Subjects were referred to impersonally as “Player 1” and “Player 2” (or “Room A” and “Room B” participants), and actions were framed strictly as “sending tokens/currency” and “returning currency.” By standardizing the physical procedures, maintaining strict incentive compatibility, and purging framing effects, the authors constructed a methodological benchmark that has served as the gold standard for laboratory replication across the global behavioral economics community.

5. Empirical Findings of the 1995 Landmark Study

5.1 Trustor Transfers and Surplus Exploration

The empirical results generated by the 1995 landmark experiment decisively rejected the subgame perfect Nash equilibrium prediction. In the baseline condition, out of 32 participating investors, 30 subjects (93.75%) transferred a non-zero quantity of capital to their anonymous counterparts ($X > 0$). Only a negligible fraction (two subjects, or 6.25%) conformed to the zero-transfer prediction of standard backward induction. The theoretical prediction of universal market failure was thus comprehensively refuted by the data.

Rather than clustering at the boundary of zero, transfers exhibited substantial dispersion across the entire domain of the action space. The mean transfer across all investors in the baseline condition was $\bar{X} =$5.16$, indicating t\hat subjects committed, on average, more than 50% of their private endowment to the uncollateralized venture. Furthermore, transfers exhibited a distinct bimodal distribution: a notable mass of participants chose to s\end intermediate amounts ($X = $5$), while a substantial contingent engaged in maximal trust, transferring their entire $X =$10$ endowment to achieve maximum social surplus.

This willingness to commit capital demonstrated that experimental subjects routinely seek to explore efficiency gains, actively defying the severe structural hazards inherent in the one-shot sequential interaction. Trustors demonstrated a systematic readiness to incur vulnerability, treating the experimental interaction not as a zero-sum contest of mutual predation, but as a collaborative opportunity to expand the total wealth of the dyad through unforced strategic vulnerability.

5.2 Trustee Responses and Return Dynamics

The behavioral patterns exhibited by Player 2 trustees were similarly at odds with standard game-theoretic predictions. If trustees had adhered to the dominant strategy dictated by pure self-interest, they would have universally retained all multiplied funds, resulting in $Y = 0$ across all positive transfers. In stark contrast to this prediction, trustees returned substantial quantities of currency to the investors, demonstrating that positive reciprocity is an active, potent behavioral driver.

Across the baseline condition, the mean return transfer was $\bar{Y} =$4.66$. When evaluated in the aggregate, the average net return to the investor was close to the break-even point: investors, on average, received back approximately what they transferred, transforming the aggregate act of trust into a roughly zero-profit proposition in financial terms, while trustees captured the lion’s share of the generated surplus. However, evaluating the data solely through aggregate averages obscures profound, systematic behavioral heterogeneity among the trustees:

• Pure Free-Riders: A significant minority of trustees acted in strict accordance with the neoclassical model, returning $Y = 0$ despite receiving substantial transfers (e.g., retaining the entire $30 generated by a$10 investment).
• Contractarian/Reciprocal Actors: A large plurality of trustees adhered to a baseline reciprocal norm, returning an amount exactly equal to the investor’s original transfer ($Y = X$), ensuring that the investor suffered no net financial harm from their trust.
• Surplus-Sharing Egalitarians: A substantial contingent of trustees engaged in strong equity preservation, returning amounts that significantly exceeded the initial transfer ($Y > X$). In many cases, these trustees divided the net surplus $(k-1)X$ evenly, returning $Y = 2X$ so that both parties enjoyed an identical net gain from the exchange.

Crucially, the authors documented a statistically significant positive correlation between the transfer amount $X$ and the absolute magnitude of the return transfer $Y$. Investors who transferred higher amounts of capital systematically elicited larger absolute dollar returns from their counterparts. This empirical regularity provided unequivocal proof of contingent, positive reciprocity: trustees calibrated their costly returns directly to the observed generosity and vulnerability demonstrated by the trustor.

5.3 Comparative Analysis: Baseline versus Social History

The comparative analysis between the Baseline treatment and the Social History treatment yielded profound insights into the resilience and institutional transmission of social norms. When a new cohort of subjects was exposed to the full historical record of the baseline sessions—revealing both the substantial generosity of investors and the opportunistic defection of the subset of free-riding trustees—economic theory predicted that the observable presence of zero-returns would trigger a rapid unravelling toward the zero-trust equilibrium.

The empirical results directly contradicted this unravelling hypothesis. The presence of social history did not extinguish trust or reciprocity; rather, it stabilized and focused normative expectations. In the Social History treatment (comprising 24 pairs), average transfers by investors remained robust at $\bar{X} =$5.36$, slightly higher than the baseline average. The proportion of subjects transferring non-zero sums remained exceptionally high, with investors concentrating their choices heavily on focal points, particularly the half-endowment ($X = $5$) and full-endowment ($X =$10$) allocations.

On the trustee side, return transfers in the Social History treatment showed an average of $\bar{Y} =$5.29$. The positive relationship between$X$ and $Y$ became even more pronounced, with trustees displaying greater consistency in rewarding large transfers. The historical data did not catalyze a race to the bottom; instead, it established a descriptive social norm, signaling to investors that trust was a recognized social practice and warning trustees that reciprocal fulfillment was an established behavioral standard within the experimental community. Berg, Dickhaut, and McCabe demonstrated that social history transforms an isolated interaction into an embedded cultural exchange, proving that social norms can maintain cooperative equilibria even when formal enforcement institutions are entirely absent.

6. The Nexus Between the Investment Game and Public Goods Dilemmas

6.1 Common Structural Roots: Costly Contributions to Shared Surplus

The deep theoretical nexus connecting the Berg, Dickhaut, and McCabe Investment Game to the extensive literature on public goods provision lies in their identical structural microfoundations: both paradigms model the agonizing friction between private opportunity costs and the realization of socially optimal collective surplus. In both classes of games, value creation requires an agent to surrender liquid, secure, private resources to initiate a production technology that yields an expanded aggregate payoff that the agent cannot unilaterally secure nor legally commandeer.

Consider the structural mapping between the two mechanisms. In a standard linear public goods Voluntary Contribution Mechanism (VCM), individual $i$ allocates contribution $g_i$ from endowment $w_i$ to a public pool, yielding individual payoff:

$$\pi_i = w_i – g_i + \alpha \sum_{j=1}^N g_j$$

Where $\alpha$ represents the Marginal Per Capita Return (MPCR). The social dilemma requires $frac{1}{N} < alpha < 1$. Under this condition, the private marginal return on contributing is$frac{partial pi_i}{partial g_i} = -1 + alpha < 0$, making complete free-riding ($g_i = 0$) the strictly dominant strategy for all players, despite the fact t\hat collective welfare is maximized when$g_i = w_i forall i$, because$frac{partial sum pi}{partial g_i} = -1 + Nalpha > 0$.

Now consider the Investment Game. The investor’s private marginal return on their transfer $X$, assuming standard game-theoretic backward induction where $Y^*(X) = 0$, is:

$$\frac{\partial \Pi_1}{\partial X} = -1 + \frac{\partial Y}{\partial X} = -1 + 0 = -1 < 0$$

Simultaneously, the marginal social return of the transfer is:

$$\frac{\partial (\Pi_1 + \Pi_2)}{\partial X} = -1 + k = -1 + 3 = 2 > 0$$

The mathematical and behavioral tension is identical. In both architectures, the socially optimal choice requires voluntary private loss in the service of collective surplus generation, and in both paradigms, the ultimate barrier to cooperative efficiency is the omnipresent hazard of opportunistic free-riding by secondary agents who enjoy the fruits of cooperation without bearing its generative costs.

6.2 Sequential Trust versus Simultaneous Public Contributions

While sharing common structural roots, the Investment Game diverges critically from traditional public goods games in its temporal choreography, providing profound analytical leverage over the cognitive mechanics of cooperation. Standard public goods games are almost universally implemented as simultaneous-move games. Under simultaneous play, when an agent chooses to free-ride ($g_i = 0$), their behavior may stem from several fundamentally distinct psychological motives:

• Predatory opportunism (the active desire to exploit the anticipated contributions of generous peers);
• Defensive fear / Risk aversion (the desire to contribute, paralyzed by the dread of being the solitary sucker who contributes while others free-ride);
• Cognitive confusion (a mathematical failure to comprehend the strategic incentives of the game).

The sequential architecture of the Berg, Dickhaut, and McCabe Investment Game cleanly uncouples these entangled motives by separating the interaction into chronological stages. By forcing the investor to move first, the game isolates vulnerability from reaction. The investor cannot exploit the trustee; they can only choose whether to extend trust under pure vulnerability. When the decision transitions to the trustee at Stage 2, all strategic ambiguity has been eliminated. The trustee does not face uncertainty regarding the investor’s action; the multiplied surplus is sitting physically before them, generated entirely by the investor’s prior voluntary sacrifice.

Consequently, when a trustee chooses to return zero ($Y = 0$), this act cannot be rationalized as defensive protection against peer free-riding; it is an act of pure, unadulterated exploitation. Conversely, when the trustee returns funds ($Y > 0$), this action represents unambiguous positive reciprocity. The Investment Game thus functions as an analytical prism, breaking the diffuse, ambiguous cooperation of simultaneous public goods games into its pure component wavelengths: vulnerability-enduring trust and intention-rewarding reciprocity.

6.3 Bilateral Reciprocity as the Microfoundational Engine of Public Goods

The conceptual insights gleaned from the Investment Game provide the microfoundational engine required to solve the grandest puzzle of public goods theory: the empirical prevalence of conditional cooperation. Empirical research by Urs Fischbacher, Simon Gächter, and Ernst Fehr has documented that roughly 50% of human beings in public goods environments are conditional cooperators—agents who are willing to contribute to the collective pool if, and only if, they anticipate or observe that others are contributing their fair share.

The Investment Game demonstrates the exact bilateral mechanism through which conditional cooperation functions. Large-scale public goods games are effectively networks of interwoven bilateral investment games. An individual’s decision to contribute tax revenue, preserve a shared fishery, or reduce carbon emissions is fundamentally an act of trust directed toward their fellow citizens. If the individual believes that their peers will reciprocate this collective investment by also restraining their private consumption, the public good flourishes. If, however, trust in peer reciprocity collapses, the individual reverts to defensive free-riding to protect themselves from asymmetric exploitation.

By formalizing reciprocity as an empirical reality, Berg, Dickhaut, and McCabe provided the analytical foundation for understanding why public goods dilemmas do not instantly degenerate into the zero-contribution tragedy of the commons predicted by standard theory. Trust acts as the initial catalyst that sparks contributions; reciprocity acts as the stabilizing feedback loop that sustains them. When reciprocity is robust, it transforms the public goods dilemma from an inescapable Prisoner’s Dilemma into an assurance game or coordination game, where high-efficiency cooperation emerges as a stable, decentralized behavioral equilibrium.

7. Psychological and Behavioral Mechanics Underpinning Decisions

7.1 Inequity Aversion and Distributional Preferences

To mathematically accommodate the empirical findings of the Investment Game, behavioral economists formulated theories of other-regarding preferences, most prominently models of inequity aversion. The pioneering framework developed by Ernst Fehr and Klaus Schmidt (1999) posits that individuals evaluate states of the world not merely by their absolute consumption, but by comparing their private payoffs to the payoffs realized by relevant reference peers. In a two-player game, an inequity-averse agent’s utility is formalized as:

$$U_i(\Pi_i, \Pi_j) = \Pi_i – \alpha_i \max{\Pi_j – \Pi_i, 0} – \beta_i \max{\Pi_i – \Pi_j, 0}$$

Where $\alpha_i$ parameterizes the disutility of suffering disadvantageous inequity (envy or grievance), $\beta_i$ parameterizes the disutility of enjoying advantageous inequity (guilt or compassion), with the structural parameter restriction $0 leq beta_i < 1$ and $\alpha_i \geq \beta_i$.

Applying this model to Player 2 (the trustee) in the Investment Game yields immediate explanatory power. Following an investment $X$, the trustee’s payoff is $\Pi_2 = M + kX – Y$, while the investor’s payoff is $\Pi_1 = M – X + Y$. For any non-zero return, the trustee experiences advantageous inequity whenever $\Pi_2 > \Pi_1$, which corresponds to the condition:

$$M + kX – Y > M – X + Y iff Y < \frac{(k+1)}{2}X$$

If the trustee’s parameter of advantageous inequity aversion is sufficiently high—specifically, if $\beta_2 > 0.5$—the marginal utility of returning an additional dollar to the investor becomes strictly positive:

$$\frac{\partial U_2}{\partial Y} = -1 + 2\beta_2 > 0$$

Under these conditions, the trustee’s optimal strategy is not to retain all funds ($Y = 0$), but rather to return funds until complete material equality is achieved: $\Pi_1 = \Pi_2$, which yields the exact focal return $Y^* = \frac{(k+1)}{2}X = 2X$ (when $k = 3$). A similar logic applies to the alternative equity model of Bolton and Ockenfels (2000), which assumes agents experience disutility whenever their private share of the aggregate surplus deviates from the egalitarian benchmark of $1/N$. Inequity aversion models thus provide a rigorous, consequentialist mathematical framework explaining why trustees voluntarily surrender wealth: they do so to extinguish the psychological pain of holding unfair, advantageous wealth at the expense of a benevolent peer.

7.2 Intention-Based Reciprocity and Psychological Game Theory

While inequity aversion models rely strictly on final payoff distributions, they suffer from an inherent blind spot: they are fundamentally outcome-based and fail to account for perceived intentionality. A human being does not merely evaluate what an exchange partner did; they evaluate why they did it. To formalize this dimension, behavioral economists turned to psychological game theory, originally formulated by John Geanakoplos, David Pearce, and Ennio Stacchetti (1989), and operationalized into economic reciprocity frameworks by Matthew Rabin (1993) and Martin Dufwenberg and Georg Kirchsteiger (2004).

In intention-based reciprocity models, an agent’s utility function incorporates explicit beliefs about the kindness or hostility underlying the counterpart’s choices. Kindness is measured by evaluating the choice made by an agent against the spectrum of alternative choices they could have selected. When Player 1 chooses a large transfer $X$ in the Investment Game, they intentionally surrender their guaranteed endowment $M$ to present Player 2 with an expanded choice set, actively risking personal poverty to endow Player 2 with wealth. This act is categorized mathematically as an unambiguous manifestation of high positive kindness:

$$\kappa_{12}(a_1, b_{12}) > 0$$

Where $\kappa_{12}$ is Player 1’s kindness toward Player 2, conditioned on action $a_1$ and first-order belief $b_{12}$ regarding Player 2’s subsequent response.

In psychological game theory, Player 2’s utility is enriched by a reciprocal incentive term:

$$U_2 = \Pi_2 + Y_{21} \cdot \kappa_{12} \cdot \lambda_{212}$$

Where $Y_{21}$ is an individual reciprocal sensitivity parameter, and $\lambda_{212}$ represents Player 2’s kindness toward Player 1. If Player 1’s intention is perceived as deeply benevolent ($\kappa_{12} > 0$), Player 2 can maximize their overall psychological utility only by selecting an action that is also deeply kind ($\lambda_{212} > 0$), which requires returning a substantial financial payoff $Y > X$. Conversely, if Player 1 is perceived to have acted stingily or insultingly (such as sending $X = 1$), Player 2 perceives this not as genuine kindness, but as an opportunistic or contemptuous provocation, leading to $\kappa_{12} \leq 0$, which rationalizes punitive zero-returns ($Y = 0$). Intention-based models explain why identical financial outcomes generate radically different behavioral responses depending upon whether they were produced by intentional human agency or stochastic nature.

7.3 Guilt Aversion and Second-Order Beliefs

A third powerful psychological mechanism governing behavior in the Investment Game is guilt aversion, formalized in the economic literature by Pierpaolo Battigalli and Martin Dufwenberg (2007). Guilt aversion posits that human agents experience severe intrinsic disutility when they let down the expectations of others—specifically, when their action causes another player’s realized payoff to fall below that player’s subjective expectation.

In this framework, the trustee’s decision-making process is mediated by their second-order beliefs. Let $\alpha_1$ represent Player 1’s first-order belief regarding the fraction of the transfer that Player 2 will return: $\alpha_1 = \mathbb{E}_1[Y]$. Guilt aversion asserts that Player 2 forms a second-order belief, denoted $\beta_2$, which represents Player 2’s estimation of what Player 1 expects to receive back:

$$\beta_2 = \mathbb{E}_2[\mathbb{E}_1[Y]]$$

If Player 2 chooses a return transfer $Y$ that falls below this inferred expectation ($Y < beta_2$), Player 2 induces psychological disappointment in Player 1. The resulting guilt experienced by Player 2 is modeled as a linear psychological penalty:

$$U_2(Y) = \Pi_2(Y) – \gamma_2 \max{\beta_2 – Y, 0}$$

Where $\gamma_2 \geq 0$ is Player 2’s personal guilt sensitivity coefficient. If $\gamma_2 > 1$, the marginal psychological cost of disappointing the investor strictly exceeds the marginal material benefit of keeping the cash ($\frac{\partial U_2}{\partial Y} = -1 + \gamma_2 > 0$). Under these preferences, the trustee’s optimal behavioral response is to set $Y^* = \beta_2$, matching their return exactly to the investor’s perceived expectations.

Extensive laboratory research utilizing belief-elicitation procedures confirms this mechanism. Econometric analyses demonstrate a powerful, positive, and statistically robust correlation between a trustee’s second-order beliefs and their actual return allocations: trustees who believe that their investors expect large returns systematically remit large returns. This finding reveals that trust functions as a self-fulfilling psychological contract: by openly demonstrating vulnerability, the trustor elevates the trustee’s second-order beliefs, generating a powerful barrier of anticipated guilt that shields the investor from opportunistic exploitation.

8. Methodological Variations and Cross-Cultural Replications

8.1 Cross-Cultural Implementations and Social Capital

The universal applicability of the findings established by Berg, Dickhaut, and McCabe has been rigorously evaluated through worldwide cross-cultural replications. These investigations sought to determine whether trust and reciprocity are biologically hardwired human universals or variable social constructs shaped by distinct cultural environments, historical institutions, and varying degrees of social capital.

In a series of landmark cross-cultural studies led by Joseph Henrich, Robert Boyd, Samuel Bowles, Colin Camerer, Ernst Fehr, and Herbert Gintis (2001, 2004), the Investment Game and related experimental paradigms were deployed across fifteen diverse, small-scale societies spanning South America, Africa, and Oceania, ranging from nomadic foragers and pastoralists to sedentary subsistence agriculturalists. The cross-cultural data revealed two profound patterns:

• The canonical model of pure, selfish Homo economicus failed universally across every single human culture studied. In no society did subjects consistently play the zero-trust subgame perfect Nash equilibrium.
• Behavioral variance across cultures was dramatically larger than variance observed across diverse demographic groups within modern Western industrial societies.

Henrich and colleagues discovered that cross-societal variations in investment transfers and reciprocal returns were strongly predicted by two structural socio-economic variables: the market integration of the society (the extent to which daily subsistence depends upon commercial market exchange) and the payoffs to cooperation (the degree to which economic survival relies upon collective, large-scale public goods endeavors, such as collective whale hunting among the Lamalera of Indonesia). Societies with high market integration and substantial reliance on collective public goods exhibited the highest levels of trust and reciprocity. Furthermore, studies linking laboratory trust games to real-world macro indicators consistently show that laboratory trust measures correlate robustly with national levels of social capital, judicial efficiency, rule of law, and gross domestic product per capita, proving that the behavioral micro-mechanisms captured by the Investment Game constitute fundamental drivers of macroeconomic development.

8.2 Strategy Method versus Direct Response Protocols

A vital methodological debate arising from the Investment Game concerns the experimental elicitation technique: the comparison between the Direct Response Method (used in Berg, Dickhaut, and McCabe’s original 1995 paper) and the Strategy Method (pioneered by Reinhard Selten, 1967). In the Direct Response protocol, the trustee sits in waiting until the investor makes an authentic, binding choice $X$; the trustee is then informed of the realized transfer and selects a single, contextualized return value $Y$. In the Strategy Method, the trustee is presented with a hypothetical schedule, required to specify their contingent return transfer $Y(X)$ for every possible transfer that Player 1 could theoretically make, before learning the actual choice made by Player 1. Once Player 1 acts, Player 2’s pre-committed response for that specific branch is automatically executed by the experimenter.

The Strategy Method provides a massive econometric advantage: it generates complete behavioral trajectories across the entire strategy space, eliciting responses even for off-equilibrium nodes (such as low or extreme transfers) that are rarely chosen in direct play. However, critics have vigorously argued that the Strategy Method introduces critical cognitive and emotional distortions:

• “Cold State” vs. “Hot State” Decision Making: The Strategy Method forces the trustee into an analytical, detached, “cold” cognitive frame, potentially dampening the visceral emotional reactions of genuine gratitude or moral obligation ignited when receiving an authentic, realized act of trust in the “hot” state of the Direct Response method.
• Cognitive Load and Framing: Presenting an exhaustive contingent table may artificially frame the interaction as an explicit mathematical function, priming subjects toward linear or proportional return strategies that they would not naturally conceptualize.

Extensive methodological meta-analyses (e.g., Brandts and Charness, 2011) comparing both elicitation modes in trust and public goods environments reveal that while the Strategy Method occasionally yields slightly lower absolute transfer and return levels due to the mitigation of visceral emotional states, the substantive behavioral patterns—specifically the positive correlation between transfer size and reciprocal return—remain remarkably robust across both protocols. The core insights established by Berg, Dickhaut, and McCabe are thus structurally resilient to the procedural mechanics of elicitation.

8.3 Repeated Interactions, Reputation, and Partner Matching

Although Berg, Dickhaut, and McCabe explicitly constructed their experiment as a strictly anonymous, one-shot game to isolate non-strategic trust, a vital strand of subsequent research has integrated their framework into repeated-interaction environments. These studies analyze how intrinsic reciprocity interfaces with strategic reputation-building, evaluating the mechanics of cooperation under Partner matching (fixed dyads interacting over repeated periods) versus Stranger matching (random re-pairing after each round to simulate repeated play while preserving one-shot anonymity).

The introduction of repeated interactions unleashes dynamic strategic incentives formalized by the folk theorem of repeated games. In a repeated Partner setup with an indefinite or unknown horizon, Player 2’s incentive to reciprocate is dramatically fortified by the shadow of the future: defection today ($Y_t = 0$) triggers severe retaliation tomorrow ($X_{t+1} = 0$), threatening the trustee’s future surplus streams. Consequently, investments and returns in repeated Partner games systematically soar to near-maximal efficiency levels ($X \approx M$, $Y > X$).

However, when the interaction is conducted across a known, finite horizon ($T$ rounds), the backward induction paradox re-emerges with theoretical fury. In round $T$, the game degenerates into a one-shot interaction; Player 2 should defect ($Y_T = 0$), which implies Player 1 should send nothing ($X_T = 0$). By backward induction, trust should collapse across all prior rounds $T-1, T-2, dots, 1$. Yet, empirical implementations reveal persistent cooperation throughout the early and intermediate rounds, followed by an abrupt “end-game collapse” in the penultimate or final round. This end-game dynamic proves that human actors fluidly synthesize strategic reputation management with intrinsic social preferences: subjects strategically deploy the appearance of trustworthiness throughout the multi-period relationship, yet the existence of non-zero returns in the final terminal round confirms that an irreducible core of intrinsic, non-strategic reciprocity persists even when the shadow of the future has completely dissolved.

9. Institutional Design and Policy Applications for Public Goods Provision

9.1 Designing Reciprocity-Based Institutional Mechanisms

The empirical reality that human societies are populated by a blend of self-interested free-riders and conditionally reciprocal actors carries profound implications for institutional design and contract theory. Classical economic policy prescriptions, rooted exclusively in the assumptions of Homo economicus, traditionally mandate that collective action problems and public goods provision must be governed by complete formal contracts, intensive third-party surveillance, and high-powered monetary sanctions designed to make defection economically irrational.

The insights of the Investment Game reveal that this heavy-handed approach can be profoundly counterproductive due to the phenomenon of institutional crowding-out. Seminal research by Ernst Fehr, Simon Gächter, and Bruno Frey demonstrates that introducing explicit, formal economic penalties into environments previously governed by informal norms can actively erode intrinsic social preferences. When a principal introduces an intrusive monitoring regime or a modest monetary fine for malfeasance, the psychological frame of the interaction shifts radically:

• The moral dimension of the transaction is extinguished; the interaction is cognitively recategorized from an ethical social relationship based on reciprocal honor into a purely transactional market exchange.
• The presence of external surveillance signals distrust, generating an insult effect that systematically diminishes the agent’s intrinsic motivation to act fairly.
• Modest fines are often treated simply as prices; agents happily pay the penalty to purchase the right to exploit the collective good guilt-free.

Consequently, sophisticated institutional design must move beyond crude command-and-control regimes. Optimal institutional architectures leverage reciprocity-based mechanisms: creating organizational environments that signal trust, granting agents sufficient autonomy to demonstrate voluntary benevolence, and employing peer-led, informal, or decentralized sanctioning mechanisms that target bad-faith actors without smothering the intrinsic moral motivations of the conditionally cooperative majority.

9.2 Information Transparency and Public Accountability

The findings derived from Berg, Dickhaut, and McCabe’s Social History condition provide a blueprint for deploying information transparency as an institutional tool to resolve public goods challenges. The 1995 study demonstrated that exposing agents to aggregate behavioral data can establish durable descriptive norms that anchor expectations and coordinate cooperative actions, even in the absence of material enforcement.

In modern public policy and corporate governance, this insight is leveraged through the deployment of non-coercive “nudges” and public accountability architectures. In environmental preservation, water conservation, and municipal energy usage, traditional economic incentives (such as Pigouvian taxes or consumption caps) frequently face intense political resistance or incur massive monitoring overheads. By contrast, public feedback mechanisms that expose individuals to descriptive social norms—such as reporting an individual’s energy consumption alongside the average consumption of their most efficient neighbors—consistently trigger substantial, voluntary reductions in resource extraction.

Similarly, in organizational management and public resource stewardship, transparency reporting transforms diffuse collective goods into visible arrays of bilateral accountabilities. When public agencies publish detailed distributions of peer contributions to collective research, charitable funds, or municipal investments, they trigger the guilt-aversion and intention-based reciprocity mechanisms documented in the laboratory: agents conform to the demonstrated standard of social contribution to avoid the severe psychological costs of public shame and personal guilt. Social history, when systematically deployed, functions as an institutional coordination device that stabilizes collective contributions at high-efficiency equilibria.

9.3 Bridging Micro-Bilateral Trust to Macro-Level Public Goods

The fundamental challenge confronting modern nation-states is scaling micro-level bilateral trust—which evolved in ancestral, small-scale tribal environments characterized by intimate face-to-face interactions—into macro-level trust capable of sustaining anonymous, complex, large-scale public goods. Every vital societal public good, from progressive taxation and public healthcare systems to public education, macroeconomic infrastructure, and climate mitigation strategies, represents a generalized, multi-agent Investment Game played between millions of anonymous citizens and the state.

When an individual citizen pays their income taxes, they are effectively enacting the role of Player 1 in an expanded Investment Game: they surrender a substantial portion of their private endowment to the sovereign state without an immediate, individual quid pro quo, under the expectation that the political collective will act as a trustworthy Player 2. If the state reciprocates this citizen investment by efficiently translating fiscal revenues into transparent, high-quality public infrastructure, impartial judicial institutions, and universal safety nets, the social contract flourishes. Citizens observe that their trust is honored, positive second-order beliefs are reinforced, and tax compliance remains exceptionally high as an expression of civic reciprocity.

Conversely, if political elites or sovereign institutions behave like predatory, free-riding trustees—embezzling resources, delivering decrepit public services, or granting asymmetric exemptions to favored oligarchic factions—the macroeconomic public goods game rapidly collapses. Citizens perceive that their vulnerability is being systematically exploited, triggering an unravelling of civic cooperation. The outcome is rampant tax evasion, the expansion of informal shadow economies, widespread corruption, and institutional sclerosis. Cross-national empirical data confirms that countries exhibiting low generalized interpersonal and institutional trust struggle continuously with fiscal degradation and public under-investment. Scaling bilateral reciprocity into institutionalized civic trust is therefore the central prerequisite for modern economic prosperity and sovereign stability.

10. Neuroeconomic and Evolutionary Foundations of the Investment Game

10.1 Neural Correlates of Trust and Reciprocity

The integration of the Investment Game into the field of neuroeconomics, accelerated by pioneering collaborations involving Kevin McCabe, Vernon Smith, Colin Camerer, and Read Montague, has unveiled the functional neural circuitry orchestrating trust and reciprocity within the human brain. Utilizing functional Magnetic Resonance Imaging (fMRI), researchers have visualized the metabolic blood-oxygen-level-dependent (BOLD) responses of subjects engaged in authentic, sequential investment decisions.

When an investor evaluates whether to transfer capital to a human counterpart, neuroimaging reveals intense activation within the ventromedial prefrontal cortex (vmPFC) and the orbitofrontal cortex (OFC)—areas fundamentally involved in the integration of emotional value, risk assessment, and social preference encoding. Simultaneously, making an investment choice engages the temporoparietal junction (TPJ), a neural locus critical for Theory of Mind (ToM) processing, reflecting the investor’s active, cognitive simulation of the trustee’s mental states, intentions, and moral commitments.

On the trustee’s side, receiving a transfer and deciding whether to return capital recruits the brain’s primary dopaminergic reward pathways: the ventral striatum, including the nucleus accumbens and the caudate nucleus. In groundbreaking hyperscanning fMRI experiments conducted by King-Casas et al. (2005), researchers imaged both trustor and trustee brains simultaneously across multi-round investment games. They discovered that when a trustee reciprocates an investment, the caudate nucleus registers a pronounced metabolic surge, demonstrating that the human brain processes reciprocal fairness not as an agonizing, costly chore, but as an intrinsically rewarding hedonic experience. Furthermore, as the relationship developed across rounds, this neural reward signal in the trustee migrated chronologically forward in time: rather than firing only after the reciprocal act was consummated, the caudate began firing as an anticipatory signal upon simply perceiving the investor’s trust, demonstrating the physiological encoding of counterpart reputation and emerging social affinity.

10.2 Biochemical and Hormonal Drivers of Trust

Beyond neuroimaging, behavioral endocrinologists have isolated specific biochemical modulators that regulate the psychological threshold for extending trust and honoring commitments, most notably the neuropeptide oxytocin. Synthesized in the hypothalamus and released both into the central nervous system and the bloodstream, oxytocin has long been implicated in mammalian pair-bonding, maternal care, and social attachment.

In a watershed pharmacological study published in Nature, Michael Kosfeld, Markus Heinrichs, Paul Zak, Urs Fischbacher, and Ernst Fehr (2005) administered exogenous intranasal oxytocin (or an identical placebo spray) to human subjects prior to playing the standard Berg, Dickhaut, and McCabe Investment Game. The empirical results were striking: subjects who received a single dose of oxytocin displayed significantly higher trust transfers than subjects in the control group. In the oxytocin condition, 44% of investors exhibited maximal trust ($X = M$), compared to only 21% in the placebo cohort. Crucially, the authors implemented a rigorous control experiment: oxytocin had zero effect on subjects’ willingness to take financial risks in a functionally identical Risky Dictator Game where the counterpart was replaced by a random lottery generator.

This biochemical dissociation demonstrated that oxytocin does not operate as a blunt chemical stimulant of generic risk-seeking behavior; rather, it specifically suppresses social fear and mitigates betrayal aversion. By dampening amygdala reactivity to social vulnerability, oxytocin permits human agents to bridge the gap of interpersonal suspicion, facilitating the emergence of collaborative investment. Conversely, physical exploitation of trust is accompanied by surges in the stress hormone cortisol and heightened activation in the anterior insula, a neural hub associated with visceral disgust, physical pain, and the acute psychological indignation ignited by unfair, exploitative interactions.

10.3 Evolutionary Dynamics of Reciprocal Altruism

The existence of costly trust and reciprocity in one-shot, anonymous laboratory settings presents a profound conundrum for Darwinian natural selection. Under classic evolutionary theory, any genetic lineage that systematically surrenders personal reproductive fitness or valuable caloric resources to benefit an unrelated, anonymous competitor should be relentlessly outcompeted and purged by natural selection, ensuring the absolute evolutionary supremacy of selfish, free-riding genotypes.

The solution to this paradox rests upon the evolutionary mechanisms of reciprocal altruism (Robert Trivers, 1971) and cultural group selection (Boyd and Richerson, 2005; Bowles and Gintis, 2011). Throughout Pleistocene ancestral environments, humans evolved in small, tightly knit hunter-gatherer bands where encounters with strangers were vanishingly rare, and completely anonymous, one-shot interactions were virtually non-existent. In such an ecological setting, almost every social interaction was characterized by repeated contact, kin relations, and comprehensive gossip-based reputational monitoring.

Under these ancestral conditions, a psychological architecture configured to default to conditional cooperation and reciprocal punishment was profoundly adaptive. Developing a heuristics-based cognitive default—an internalized behavioral rule to “extend trust, reward generosity, and punish betrayal”—maximized long-term inclusive fitness by guaranteeing continuous inclusion in collective hunting, food-sharing, and mutual defense alliances. When modern human subjects are placed inside the ultra-sterile, artificial conditions of a 1995 double-blind laboratory experiment, their evolutionary psychology does not instantly switch off. The neural and emotional heuristics forged over hundreds of millennia continue to govern behavior, leading subjects to treat anonymous peers not as disposable game-theoretic abstractions, but as potential partners in the ancient human enterprise of reciprocal survival.

11. Methodological Critiques, Confounds, and Open Debates

11.1 Disentangling Trust from Pure Altruism: The Cox Decomposition

Despite its monumental influence, the original Berg, Dickhaut, and McCabe experimental design faced significant methodological scrutiny regarding identification. The most rigorous and influential critique was formulated by James C. Cox in his seminal 2004 paper, “Trust and Reciprocity: Implications of Game Playing Risk Analysis.” Cox argued that the canonical Investment Game systematically conflates trust with pure, unconditional altruism on the part of the investor, and conflates reciprocity with unconditional warm-glow or baseline inequality aversion on the part of the trustee.

Cox demonstrated that an investor who transfers $X > 0$ might not be expecting a return at all; they may simply be an altruist who derives utility from transferring wealth to someone who values it, exactly as individuals do in unilateral Dictator Games. Similarly, a trustee who returns $Y > 0$ might do so purely out of general benevolence or an aversion to being materially richer than the investor, regardless of whether the investor’s transfer was motivated by trust or accidental assignment.

To definitively isolate genuine trust and reciprocity from baseline altruism, Cox developed a brilliant triadic experimental design comprising three distinct, between-subject treatments:

• Treatment A (Standard Trust Game): The baseline Berg, Dickhaut, and McCabe Investment Game, providing a composite measure of transfers and returns.
• Treatment B (Investor Dictator Control): Player 1 is given an endowment and can unilaterally transfer money to Player 2, where transfers are multiplied by three. However, Player 2 is completely passive and has no capacity to return any funds. Transfers in Treatment B measure pure altruistic giving, completely purged of any expectation of return.
• Treatment C (Trustee Dictator Control): Subjects are exogenously endowed with payoffs identical to those realized in Treatment A following an investor transfer, but subjects are informed that this initial distribution was determined entirely by the experimenter, with no voluntary sacrifice or trust initiated by Player 1. Player 2 can return money to Player 1. Returns in Treatment C measure pure distributional preferences, completely purged of reciprocal obligation toward the investor.

By subtracting the transfer levels observed in Treatment B from those in Treatment A, Cox mathematically isolated genuine, unconfounded trust. Similarly, by subtracting the return transfers observed in Treatment C from those in Treatment A, he isolated pure, contingent positive reciprocity. The empirical execution of the Cox decomposition revealed that while pure altruism and distributional preferences do account for a noticeable fraction of laboratory transfers, the differentials $(A – B)$ and $(A – C)$ remain positive, statistically significant, and large. Cox’s methodological refinement validated the qualitative conclusions of Berg, Dickhaut, and McCabe while providing a mathematically rigorous framework for isolating the distinct motivational components of social preferences.

11.2 Confounding Risk Preferences with Social Expectations

A second major methodological critique interrogates the degree to which the investor’s choice in the Investment Game evaluates relational trust versus generic risk tolerance. In a standard microeconomic portfolio allocation problem, an investor allocates capital between a safe asset (the retained endowment $M – X$) and a risky asset (the transferred capital $X$, which yields a stochastic return $Y$). Under this framing, the transfer $X$ could be fully explained by standard Expected Utility theory: if the investor believes that the expected return $\mathbb{E}[Y | X] > X$ and their coefficient of risk aversion is sufficiently low, they will invest, purely as a mathematical gamble.

To resolve whether trust is reducible to generic financial risk preferences, experimentalists engineered comparative designs pitting the standard Investment Game against a “Trust Lottery” or “Risky Choice Game.” In these lottery treatments, subjects make an identical investment decision, but the return transfer is determined not by a human counterpart, but by a mechanical randomizer (such as a roulette wheel or computerized bingo blower) calibrated to match the exact historical probability distribution of human trustee returns. If the transfer $X$ is governed purely by financial risk aversion under subjective probabilities, transfers in the human Trust Game and the mechanical Lottery Game should be statistically indistinguishable.

The empirical findings across dozens of independent studies (e.g., Bohnet and Zeckhauser, 2004; Bohnet et al., 2008) reveal a massive, systematic deviation: subjects transfer significantly more money in the mechanical lottery condition than in the human trust condition. This gap has been christened betrayal aversion. Human beings demand a substantial risk premium to expose themselves to the hazard of voluntary, intentional human betrayal compared to facing impersonal, mechanical risks of identical mathematical probability. Trust, therefore, carries a profound psychological overhead that cannot be captured by standard risk metrics. It represents a specialized cognitive and emotional calculus where agents explicitly weigh the social vulnerability of interpersonal relationships against the potential rewards of collaborative value creation.

11.3 External Validity and Ecological Boundaries

A persistent debate surrounding the Investment Game, and experimental economics broadly, centers on the question of external validity: do the choices made by university undergraduate students passing envelopes or clicking buttons inside an artificial laboratory environment predict real-world economic behavior? Critics, notably Steven Levitt and John List (2007), have argued that laboratory experiments introduce profound selection biases, artificial scrutiny, unrepresentative demographic samples, and low stakes that limit generalizability to natural market environments.

Scholars have systematically confronted these critiques along three key empirical dimensions:

• Stake Size Effects: Orthodox theorists argued that subjects would abandon trust and reciprocity if real-world stakes were escalated to substantial levels. In response, Cameron (1999), Fehr, Fischbacher, and Tougareva, and others replicated the Investment Game in developing nations (such as Indonesia and Russia) where the experimental endowments equalled several weeks or even months of local wages. The empirical results were definitive: while transfers became slightly more cautious, the structural patterns of trust and positive reciprocity persisted unabated. Multiplied surplus generation and proportional reciprocal returns remained robust even when choices involved substantial personal income.
• Demographic and Field Generalizability: Replications conducting trust games with non-student populations—including microfinance borrowers in Peru (Karlan, 2005), commercial bank managers, and small business owners—consistently demonstrate that laboratory measures of trust correlate with real-world repayment rates, informal credit access, and contractual reliability.
• Contextual Framing: While sterile laboratory instructions deliberately purge social cues, real-world institutions operate within rich contextual frames. Research demonstrates that introducing naturalistic framing (e.g., describing the game as a supplier-buyer contract or agricultural partnership) typically amplifies cooperative norms rather than diminishing them, proving that the laboratory setting establishes a conservative lower bound for the empirical prevalence of trust in human society.

12. Legacy and Contemporary Relevance of Berg, Dickhaut, and McCabe’s Work

12.1 Paradigm Shift in Contract Theory and Incomplete Contracts

The foundational discoveries of Berg, Dickhaut, and McCabe catalyzed an intellectual revolution within microeconomic contract theory, transforming the analytical treatment of incomplete contracts. Classical contract theory, rooted in the framework of Oliver Hart, Sanford Grossman, and John Moore, operates under the presumption that when effort, quality, or environmental contingencies are non-verifiable by a third-party court of law, bilateral trade suffers severe hold-up problems, leading to under-investment and economic inefficiency.

The empirical validation of trust and reciprocity forced contract theorists to radically expand their formal frameworks. In a landmark series of papers on “gift-exchange” and relational contracting, Ernst Fehr, Georg Kirchsteiger, and Arno Riedl demonstrated that when principals offer generous, above-market wages (analogous to Player 1’s generous transfer $X$), agents reciprocate by voluntarily delivering costly, non-verifiable effort far exceeding the minimum contractually enforceable threshold (analogous to Player 2’s return $Y$). Trust-based informal contracts routinely achieve dramatically higher economic efficiency than rigid, explicit contracts that attempt to verify every contingency through formal surveillance.

This insight fundamentally reshaped modern organizational economics. Modern corporations are no longer modeled purely as webs of legal sanctions and panoptic monitoring systems, but as social institutions whose efficiency depends critically upon internal cultures of mutual trust and reciprocity. Discretionary managerial trust empowers employees to exercise creative problem-solving and unmonitorable effort, solving collective action problems within the firm that formal contractual engineering is fundamentally powerless to resolve.

12.2 Digital Trust, Decentralized Networks, and Modern Public Goods

In the twenty-first century, the structural architecture of the Investment Game has found an unexpected and vital application within the digital economy, peer-to-peer (P2P) platforms, and cryptographic decentralized networks. Modern digital platforms—such as Airbnb, Uber, eBay, and open-source software ecosystems—are essentially scaled, automated implementations of generalized investment games. A user transacting on a decentralized network transfers capital, personal safety, or proprietary digital assets to an anonymous counterpart or autonomous smart contract, with zero physical co-presence and often across distinct sovereign jurisdictions where traditional legal recourse is impossible.

The sustainability of these massive digital public goods relies on architectures engineered directly from the insights of experimental economics:

• Reputation Protocols as Institutionalized Social History: Online rating systems, proof-of-stake mechanisms, and cryptographic reputation ledgers serve as the modern, algorithmic equivalent of Berg, Dickhaut, and McCabe’s Social History condition, continuously broadcasting the historical behavioral distributions of network actors to coordinate expectations and punish opportunistic defection.
• Algorithmic Trust vs. Social Reciprocity: The rise of Decentralized Autonomous Organizations (DAOs) and blockchain protocols highlights the profound tension between deterministic code (“trustless” systems) and intrinsic human social preferences. Empirical evidence demonstrates that even in environments governed by automated smart contracts, governance deadlocks, hard forks, and public goods under-funding (such as open-source protocol maintenance) cannot be resolved by cryptographic incentives alone; they require underlying human communities bound by shared normative trust and conditional reciprocity.
• Digital Commons and Quadratic Funding: Contemporary mechanisms for funding public goods in digital ecosystems, such as Gitcoin’s implementation of Quadratic Funding (pioneered by Glen Weyl, Vitalik Buterin, and Zoë Hitzig), explicitly rely on bilateral community donations to guide matching capital pools, utilizing the reciprocal contributions of micro-investors to dynamically identify and finance high-value societal public goods.

12.3 Synthesis: An Integrated Theory of Trust, Reciprocity, and Public Goods

Thirty years after its initial publication, the Investment Game devised by Joyce Berg, John Dickhaut, and Kevin McCabe stands as a foundational monument in the landscape of the behavioral and social sciences. By conceptualizing trust and reciprocity not as mystical moral virtues or strategic illusions, but as measurable, formal, and economically potent behaviors, their work dismantled the dogmatic hegemony of Homo economicus and unlocked the modern empirical era of social preferences.

The synthesis of three decades of intensive research reveals an integrated architecture of human social cooperation:

$$\text{Micro-Level Trust} x\rightarrow{\text{Dynamic Reciprocity}} \text{Relational Exchange} x\rightarrow{\text{Information / Norms}} \text{Macro-Level Public Goods}$$

Bilateral trust operates as the foundational microeconomic spark that initiates trade, expands efficiency horizons, and undertakes productive vulnerabilities. Reciprocity functions as the behavioral stabilizer that honors that vulnerability, suppresses opportunistic predation, and enforces fairness norms without the deadweight loss of costly legal systems. Scaled across institutions through information transparency, cultural social history, and enlightened institutional design, these dual mechanisms provide the microfoundational engine that sustains the entire matrix of modern collective action—transforming fragile human groups into resilient, cooperative societies capable of overcoming the tragedies of the commons to build enduring public goods.

Conclusion

The seminal contribution of Joyce Berg, John Dickhaut, and Kevin McCabe in their 1995 paper, “Trust, Reciprocity, and Social History,” fundamentally revolutionized our understanding of human economic agency. By designing an austere yet powerful experimental architecture, they demonstrated that uncollateralized economic exchange does not collapse into opportunistic ruin as predicted by standard non-cooperative game theory. Instead, their work provided empirical proof that human beings are fundamentally endowed with behavioral preferences for trust and reciprocity—preferences capable of generating vast social surpluses under conditions of complete anonymity and non-repeated play.

Throughout this comprehensive investigation, we have demonstrated that the Investment Game is not an isolated laboratory curiosity; it is the elementary particle of societal collective action. From its precise mathematical mechanics and psychological foundations in inequity aversion, intention-based kindness, and guilt aversion, to its neural substrates in the striatum and prefrontal cortex, the game captures the essential dynamics of interpersonal cooperation. Furthermore, by linking the bilateral sequential architecture of the Investment Game to the multilateral simultaneous dilemmas of public goods provision, we see that large-scale collective endeavors are fundamentally networks of bilateral trust relationships mediated by conditional reciprocity.

As human society continues to navigate unprecedented collective action challenges—ranging from the preservation of global planetary commons and the equitable allocation of digital public goods to the institutional governance of algorithmic networks and decentralized finance—the lessons formulated by Berg, Dickhaut, and McCabe remain profoundly urgent. They remind us that the wealth of nations does not depend solely upon the physical machinery of markets or the coercive threat of sovereign legal sanctions. Rather, it rests upon the fragile, magnificent, and uniquely human capacity to extend our resources to the discretion of an anonymous peer, grounded in the enduring expectation that our vulnerability will be honored through reciprocal benevolence.

References

• Battigalli, P., & Dufwenberg, M. (2007). Guilt in games. American Economic Review, 97(2), 170–176. https://doi.org/10.1257/aer.97.2.170
• Berg, J., Dickhaut, J., & McCabe, K. (1995). Trust, reciprocity, and social history. Games and Economic Behavior, 10(1), 122–142. https://doi.org/10.1006/game.1995.1027
• Bohnet, I., Greig, F., Herrmann, B., & Zeckhauser, R. (2008). Betrayal aversion: Evidence from Brazil, China, Oman, Switzerland, and the United States. American Economic Review, 98(1), 294–310. https://doi.org/10.1257/aer.98.1.294
• Bohnet, I., & Zeckhauser, R. (2004). Trust, risk and betrayal. Journal of Economic Behavior & Organization, 55(4), 467–484. https://doi.org/10.1016/j.jebo.2003.11.004
• Bolton, G. E., & Ockenfels, A. (2000). ERC: A theory of equity, reciprocity, and competition. American Economic Review, 90(1), 166–193. https://doi.org/10.1257/aer.90.1.166
• Bowles, S., & Gintis, H. (2011). A cooperative species: Human reciprocity and its evolution. Princeton University Press. https://press.princeton.edu/books/hardcover/9780691151250/a-cooperative-species
• Boyd, R., & Richerson, P. J. (2005). The origin and evolution of cultures. Oxford University Press. https://doi.org/10.1093/oso/9780195165241.001.0001
• Brandts, J., & Charness, G. (2011). The strategy versus the direct-response method: A first survey of experimental comparisons. Experimental Economics, 14(3), 375–398. https://doi.org/10.1007/s10683-011-9272-x
• Cameron, L. A. (1999). Raising the stakes in the ultimatum game: Experimental evidence from Indonesia. Economic Inquiry, 37(1), 47–59. https://doi.org/10.1111/j.1465-7295.1999.tb01415.x
• Charness, G., & Rabin, M. (2002). Understanding social preferences with simple tests. Quarterly Journal of Economics, 117(3), 817–869. https://doi.org/10.1162/003355302760193904
• Cox, J. C. (2004). How to identify trust and reciprocity. Games and Economic Behavior, 46(2), 260–281. https://doi.org/10.1016/S0899-8256(03)00119-2
• Dufwenberg, M., & Kirchsteiger, G. (2004). A theory of sequential reciprocity. Games and Economic Behavior, 47(2), 268–298. https://doi.org/10.1016/j.geb.2003.06.003
• Fehr, E., Fischbacher, U., & Tougareva, E. (2002). Do high stakes dismantle other-regarding preferences? Evidence from Russia (Working Paper No. 112). Institute for Empirical Research in Economics, University of Zurich. https://doi.org/10.5167/uzh-51759
• Fehr, E., & Gächter, S. (2000). Cooperation and punishment in public goods experiments. American Economic Review, 90(4), 980–994. https://doi.org/10.1257/aer.90.4.980
• Fehr, E., Kirchsteiger, G., & Riedl, A. (1993). Does fairness prevent market clearing? An experimental investigation. Quarterly Journal of Economics, 108(2), 437–459. https://doi.org/10.2307/2118338
• Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition, and cooperation. Quarterly Journal of Economics, 114(3), 817–868. https://doi.org/10.1162/003355399556151
• Fischbacher, U., Gächter, S., & Fehr, E. (2001). Are people conditionally cooperative? Evidence from a public goods experiment. Economics Letters, 71(3), 397–404. https://doi.org/10.1016/S0165-1765(01)00394-9
• Forsythe, R., Horowitz, J. L., Savin, N. E., & Sefton, M. (1994). Reassessing fairness in ultimatum and dictator games. Games and Economic Behavior, 6(3), 347–369. https://doi.org/10.1006/game.1994.1021
• Frey, B. S., & Jegen, R. (2001). Motivation crowding theory. Journal of Economic Surveys, 15(5), 589–611. https://doi.org/10.1111/1467-6419.00150
• Geanakoplos, J., Pearce, D., & Stacchetti, E. (1989). Psychological games and sequential rationality. Games and Economic Behavior, 1(1), 60–79. https://doi.org/10.1016/0899-8256(89)90005-5
• Henrich, J., Boyd, R., Bowles, S., Camerer, C., Fehr, E., Gintis, H., & McElreath, R. (2001). In search of Homo economicus: Behavioral experiments in 15 small-scale societies. American Economic Review, 91(2), 73–78. https://doi.org/10.1257/aer.91.2.73
• Henrich, J., Boyd, R., Bowles, S., Camerer, C., Fehr, E., & Gintis, H. (Eds.). (2004). Foundations of human sociality: Economic experiments and ethnographic evidence from fifteen small-scale societies. Oxford University Press. https://doi.org/10.1093/0199262055.001.0001
• Hoffman, E., McCabe, K., & Smith, V. L. (1996). Social distance and other-regarding behavior in dictator games. American Economic Review, 86(3), 653–660. https://www.jstor.org/stable/2118214
• Isaac, R. M., Walker, J. M., & Thomas, S. H. (1984). Divergent evidence on free riding: An experimental examination of possible explanations. Public Choice, 43(2), 113–149. https://doi.org/10.1007/BF00140829
• Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1986). Fairness and the assumptions of economics. Journal of Business, 59(S4), S285–S300. https://www.jstor.org/stable/2352761
• Karlan, D. S. (2005). Using experimental economics to measure social capital and predict financial decisions. American Economic Review, 95(5), 1688–1699. https://doi.org/10.1257/000282805775014407
• King-Casas, B., Tomlin, D., Anen, C., Camerer, C. F., Montague, P. R., & Montague, P. R. (2005). Getting to know you: Reputation and trust in a two-person economic exchange. Science, 308(5718), 78–83. https://doi.org/10.1126/science.1108062
• Kosfeld, M., Heinrichs, M., Zak, P. J., Fischbacher, U., & Fehr, E. (2005). Oxytocin increases trust in humans. Nature, 435(7042), 673–676. https://doi.org/10.1038/nature03701
• Levitt, S. D., & List, J. A. (2007). What do laboratory experiments measuring social preferences reveal about the real world? Journal of Economic Perspectives, 21(2), 153–174. https://doi.org/10.1257/jep.21.2.153
• McCabe, K., Houser, D., Ryan, L., Smith, V., & Trouard, T. (2001). A functional imaging study of cooperation in two-person reciprocal exchange. Proceedings of the National Academy of Sciences, 98(20), 11832–11835. https://doi.org/10.1073/pnas.211415698
• Rabin, M. (1993). Incorporating fairness into game theory and economics. American Economic Review, 83(5), 1281–1302. https://www.jstor.org/stable/2117790
• Selten, R. (1967). Die Strategiemethode zur Erforschung des eingeschränkt rationalen Verhaltens im Rahmen eines Oligopolexperiments. In H. Sauermann (Ed.), Beiträge zur experimentellen Wirtschaftsforschung (pp. 136–168). J.C.B. Mohr (Paul Siebeck).
• Trivers, R. L. (1971). The evolution of reciprocal altruism. Quarterly Review of Biology, 46(1), 35–57. https://doi.org/10.1086/406755