MANAGERIAL AND DECISION ECONOMICS Manage. Decis. Econ. 27: 103–129 (2006) Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/mde.1287 Cognitive for n-person Exchange: The Evolutionary Roots of Organizational Behavior

John Toobya,*, Leda Cosmidesa and Michael E. Priceb,c a Center for , Department of Anthropology, University of California, Santa Barbara, CA 93106-3210, USA b Indiana University Workshop in Political Theory and Policy Analysis and The Santa Fe Institute, Indiana University, 513 N. Park, Bloomington, IN 47408-3895, USA c Olin School of Business, Washington University in St. Louis, Campus Box 1133, 1 Brookings Drive, St. Louis, MO, 63130-4899, USA

Organizations are composed of stable, predominantly cooperative interactions or n-person exchanges. have been engaging in n-person exchanges for a great enough period of evolutionary time that we appear to have evolved a distinct constellation of -typical mechanisms specialized to solve the adaptive problems posed by this form of social interaction. These mechanisms appear to have been evolutionarily elaborated out of the cognitive infrastructure that initially evolved for dyadic exchange. Key adaptive problems that these mechanisms are designed to solve include coordination among individuals, and defense against exploitation by free riders. Multi-individual cooperation could not have been maintained over evolutionary time if free riders reliably benefited more than contributors to collective enterprises, and so outcompeted them. As a result, humans evolved mechanisms that implement an aversion to exploitation by free riding, and a strategy of conditional cooperation, supplemented by punitive sentiment towards free riders. Because of the design of these mechanisms, how free riding is treated is a central determinant of the survival and health of cooperative organizations. The mapping of the evolved psychology of n-party exchange cooperation may contribute to the construction of a principled theoretical foundation for the understanding of behavior in organizations. Copyright # 2006 John Wiley & Sons, Ltd.

EVOLUTIONARY FORMS OF transparently based on general-purpose abilities to RATIONALITY: THE EVOLUTIONARY AND reason, pursue goals and select the highest payoffs. COGNITIVE BACKGROUND TO EXCHANGE Although collective action has a similar ability to AND MULTI-INDIVIDUAL COOPERATION improve the net welfare of participants, econo- mists and organizational theorists have long Voluntary exchange for mutual benefit improves realized that there was a problem in attempting the net welfare of participants. Consequently, to explain collective action as a parallel expression traditional economic and social science theories of rationality plus self-interest (Olson, 1965; often treat exchange simply as the straightforward Ostrom, 1998; Price et al., 2002). In n-party product of human rationality plus self-interest, interactions that produce public and common goods (that is, produce outcomes where individual effort is initially unlinked to the ability to consume *Correspondence to: Center for Evolutionary Psychology, Department of Anthropology, University of California, Santa the benefits of a joint effort), free riding provides Barbara, CA 93106-3210, USA. E-mail: [email protected] higher payoffs than contributing. As a result,

Copyright # 2006 John Wiley & Sons, Ltd. 104 J. TOOBY ET AL. rational agents}who according to standard eco- these procedures constitute functional design nomic theory will be essentially everyone}will features that led to incorporate choose not to contribute. Because of this free rider them into the human species-typical architecture? problem, it is puzzling why any rational, self- That is, what adaptive problems are these proce- interested individual would contribute at all to a dures designed to solve, and in what way do their collective action. Yet, across all known cultures, patterned outputs correspond to solutions? Evolu- throughout history, humans have commonly en- tionary psychology has been effective as a research gaged in collective actions (Ostrom, 1990; Price strategy because understanding the adaptive pro- et al., 2002; Tooby and Cosmides, forthcoming). blems faced by our ancestors provides detailed Where interactions are small in scale, joint efforts predictions about the designs}the functional are routine. How can this be? architecture}of our evolved psychological me- chanisms, guiding new empirical initiatives. Consequently, taking an evolutionary psycho- Locating Rationality, Choice, and Organizational logical approach offers a new framework for the Behavior in an Evolutionary Psychological study of rationality, choice, and organizational Framework behavior that differs in certain respects from more Although contributing to collective actions ap- traditional approaches. Because reliably develop- pears to violate standard rationality and the theory ing species-typical mechanisms are the product of of choice behavior that emerges from it, rationality evolution, the functional logic they embody will be itself appears less straightforward when the a species of evolutionary rationality}what we attempt is made to locate formal principles of have called ecological rationality (Tooby and rationality and economic choice in the causal Cosmides, 1992b). Ecological rationality differs matrix of the physical world. First, rationality and from ordinary concepts of rationality and func- choice are not, of course, mechanism-free, but are tionality in several ways: real solely to the extent they are embodied in the First, the functional product that evolved information-processing architecture of neural pro- mechanisms are designed to produce is not utility grams in the human brain. Economic and organi- maximization, social welfare or the general ability zational behaviors do not flow from disembodied to realize goals (although it may overlap with these principles, but are the computed outputs of under many circumstances). Instead, the mechan- structured cognitive mechanisms incarnated in isms were designed (in interaction with the other brain organization. So, cognitive science is one mechanisms in the architecture) to produce out- foundation needed to understand human econom- puts that typically promoted genic fitness under ic and organizational decision-making. Second, ancestral conditions (that is, that increased the species-typical mechanisms of reasoning and frequency of the mechanisms’ genetic basis in choice, in whatever form they exist, came into subsequent generations relative to pre-existing or being because they (or their developmental bases) mutational alternatives). This definition of func- were produced by the evolutionary process. So, tionality often departs from both formalized nor- evolutionary biology provides a second founda- mative theories of rationality and common sense tion for understanding human decision-making. intuitions about functionality. For example, me- Evolutionary psychology unites the two projects chanisms involved in dyadic or n-person exchange of evolutionary functionalism and cognitive may, by design, cause choices that do not maximize science into a single integrated research program absolute payoffs, violating standard rational prin- (Tooby and Cosmides, 1992a). It explicitly uses ciples (e.g. Hoffman et al., 1998; Price et al., 2002). knowledge of what natural selection would have Similarly, mechanisms underlying jealousy violate favored during to guide the common sense notions of functionality. Jealousy empirical mapping of the computational architec- often causes desperate unhappiness for the in- ture of the human mind/brain. Thus, it asks two dividual in which it is activated, as well as for its sets of interrelated, complementary, and mutually targets}a byproduct of the fact that its function is illuminating questions. First, what are the actual to spread the genetic basis of jealousy at the computational procedures that the human mind/ expense of the alternative design (indifference to brain embodies in its reasoning, emotion, and the sexual behavior of one’s mate), regardless of its motivational systems? Second, which aspects of impact on the jealous individual’s happiness.

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 105

Second, these mechanisms were designed to be behavior or donating blood) will come to be evolutionarily functional within the ecological recognized as engineering byproducts of these (causal) structure of the ancestral world humans functional designs when they are activated outside evolved in, but not necessarily in environments of the ancestral envelope of conditions for which whose structure departs from ancestral conditions. they were designed. On this view, modern ‘irrationality’ will commonly be the expression of ancestral functionality: A A Surprise: Exchange is not Produced by General taste for salt evolved under conditions where Rationality but by Evolved Neurocomputational salt}a necessary nutrient}was so chronically Programs Functionally Specialized for scarce that modern humans ‘irrationally’ over- Exchange Interactions consume it now that it is abundant. Third, our cognitive architecture system has not One of the first applications of the evolutionary been designed by evolution to reach with equal psychological research program was to the phe- efficiency any goal that arbitrary preferences might nomenon of dyadic exchange (i.e. where two nominate. Rather, our species-typical neurocom- parties deliver benefits to each other, each delivery putational architecture evolved to solve particular being made conditional on the other). The primary families of adaptive problem that recurred fre- questions addressed were: What do the actual quently enough to select for mechanisms to solve computational procedures underlying exchange in them. Hence, the design of specific problem- humans look like, and how do their design features solving systems is matched to the recurrent reflect solutions to the adaptive problems imposed structure of ancestrally significant adaptive pro- by the recurrent structure of ancestral social blems just as specific keys are machined to fit exchanges? Perhaps the largest surprise to emerge particular locks. from this research was the finding that exchange, How does this apply to understanding multi- despite being precisely the kind of thing that individual cooperation and collective action? We general rationality would guide rational agents to think that human evolutionary history has engage in, is not after all produced by a general- equipped the human mind with specialized psy- purpose set of reasoning abilities. That is, the chological adaptations designed to realize gains in traditional view that exchange is the expression of trade that occur both in 2-party exchanges and in general economic rationality turns out to be false. n-party exchanges, including collective actions. We Instead, exchange is produced by evolved reason- believe that the specific characteristics of these ing specializations}social exchange procedur- mechanisms (e.g. cheater detection circuits) reflect es}tailored by natural selection to solve the the ancestrally recurrent structure of these adap- computational problems specific to social ex- tive problems (the existence of payoffs to cheat- change (Cosmides, 1989; Cosmides and Tooby, ing), just as parts of the key shape complement 1989, 1992, 2005; Hoffman et al., 1998). What detailed parts of the lock. We think the under- leads to this conclusion? Exchange depends on standing of collective action and organizational conditionally delivered behavior (e.g. I will do X if behavior can be improved by exploring the you do Y), and so necessarily requires conditional properties of this evolved psychology. We expect reasoning for its regulation. In theory, of course, that emerging evolutionarily psychological the- this could be accomplished by general reasoning ories of n-person exchange and collective action abilities}the traditional assumption. Yet when will replace theories of standard economic ration- methods were developed to study conditional ality by explaining their puzzles. In particular, we reasoning performance in humans, subjects turned think that violations of standard economic ration- out to be very poor at detecting potential ality that are often found in situations studied by violations of conditional rules across a very broad researchers in organizational behavior, behavioral range of familiar and unfamiliar contents (see economics, and anthropology are expressions of review in Cosmides, 1989). One could not attribute well-engineered, ecologically rational mechanisms the human ability to engage in exchange to a producing outputs that would have been advanta- general ability to reason about conditionals. geous in ancestrally typical conditions. Indeed, we Based on an adaptationist analysis of exchange, think that many of the ‘irrational’ behavioral it was hypothesized that humans have an evolved expressions of these mechanisms (such as voting cognitive specialization for reasoning about social

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 106 J. TOOBY ET AL. exchange, including a subroutine for detecting refer to the superordinate category of intercontin- cheaters (Cosmides, 1989; Cosmides and Tooby, gent interactions involving gains in trade). 1989). This led to a number of novel and specific George Williams was, characteristically, the first predictions. For example, despite being generally to introduce the topic of reciprocation into poor at detecting potential violations of condi- evolutionary biology, perceptively identifying the tional rules, subjects should nevertheless detect key strategy as one of the conditionality of one act them readily when the rule involves social ex- of benefit delivery on the other (Williams, 1966). change and looking for violations corresponds to Trivers (1971) elaborated Williams’ insights into a looking for cheaters. Second, subjects’ reasoning far richer treatment, in the light of the already choices should correspond to the adaptively existing experimental game theory literature on correct choices laid out in a specialized logic of iterated prisoners’ dilemmas and the social psy- social exchange, even when these choices conflicted chology literature on reciprocity. This was fol- with logically correct choices. These and related lowed by more formal and systematic analyses predictions were subsequently confirmed by nu- using evolutionary game theory, such as Axelrod merous experiments conducted on subject popula- and Hamilton’s exploration of tit-for-tat and tions drawn from many societies, from Harvard Maynard Smith’s ESS analyses of reciprocation undergraduates to Amazonian hunter–horticultur- (Axelrod and Hamilton, 1981; Maynard Smith, alists (for review, see Cosmides and Tooby, 2005). 1982). One result that robustly emerges from most Indeed, evidence from cognitive neuroscience evolutionary game theory formalizations of reci- shows it is possible through brain damage to have procation is that it cannot easily evolve unless the mechanisms underlying social exchange rea- reciprocation strategies avoid cheaters. This im- soning selectively impaired, while other reasoning plicitly assumes that reciprocation strategies have abilities remain intact}something that would the capacity to detect instances of cheating not be possible if reasoning about different (noncompliance, defection). Thus, although the contents were accomplished using the same set fitness advantages of exchange are no puzzle, and of general-purpose rational mechanisms (Stone their elucidation is a ratification rather than a et al., 2002.) contribution to economics, the theoretical analysis Identifying the fitness advantages of exchange, of reciprocation highlighted an evolutionary vul- of course, was not a deep evolutionary puzzle: As nerability to the strategy that our cognitive and has been recognized at least since Adam Smith (if motivational machinery would necessarily have not for hundreds of thousands of generations), been selected to address: A reciprocation strategy, exchange increases the welfare of both parties. to be successful, must incorporate defenses against What is critical from an evolutionary perspective is being outcompeted by cheaters. This result applies whether the mechanisms that cause us to engage in not only to reciprocation, but to the more exchange have been shaped in any specific ways by encompassing category of exchange (Cosmides selection pressures that are particular to exchange and Tooby, 1989, 1992). This conclusion led to interactions (Cosmides, 1989; Cosmides and Too- the prediction and discovery that humans have a by, 1989). If exchange had turned out to be simply cognitive specialization that allows them to detect one out of an indefinitely large set of activities violations of conditional rules when the condi- made possible by general learning abilities or tional relationship involves exchange and a viola- general rationality, then an evolutionary approach tion would constitute an act of cheating (Cosmides would have had little to contribute. Of course, the and Tooby, 2005). evolutionary biology community has not widely If 2-party exchange is not caused by general addressed the superordinate category of exchange rationality, then contributing to collective action is per se (Tooby and Cosmides, 1996), but primarily not}after all}an anomaly because it departs a narrower subset of exchange interactions: alter- from general rationality. Because general ration- nating, deferred (usually implicit) exchange. (Tri- ality (as a set of real, physical mechanisms) vers, 1971) confusingly labeled this behavior seemingly does not exist, departures from it do reciprocal altruism, although it did not meet the not require explanation.1 Instead, both types of biological or common sense definition of altruism. behavior are explained by the actual designs of the We will use reciprocation to refer to alternating evolved neurocomputational programs that cause deferred exchange and exchange or reciprocity to them. But what exactly are these designs, and how

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 107 could they have emerged over evolutionary time in tive problems (Tooby et al., 2005). Thus, a fear of the face of potentially fitter competitive challenges snakes is not the product of a general rational by alternative designs? On this view, both 2-person system for avoiding negative payoffs, but rather an and n-person exchange behavior appear (at first evolved system with evolved proprietary represen- examination) to be anomalous from an evolu- tations (snake) and motivational dispositions tionary perspective because both suffer from (fear/avoidance). This system is ecologically ra- parallel vulnerabilities to evolved counterstrategies tional in that its specialized motivational circuits that should have outcompeted them: Cheaters and motivate behavior that was ancestrally functio- free riders both take benefits without making nal}snake avoidance}even though it may some- contributions, and so would have outcompeted times lead to unnecessary (‘irrational’) avoidance from the beginning simple first-order designs for 2- of some harmless snakes (i.e. it is not perfect). The party or n-party unconditional ‘exchange’ (benefit key claim is that for motivation to direct behavior delivery). Consequently, what an evolutionary adaptively, it requires guidance systems equipped perspective predicts instead is that the mechanisms with evolved representational elements capable of that evolved must necessarily embody more correctly specifying the targets and intensities of complex, conditional exchange strategies whose the motivation. Examples of such target-specifying computational properties defend them against ‘innate ideas’ that reliably develop within the outcompetition by cheaters (in 2-party exchange) n-person exchange motivational system include and free-riders (in n-person exchanges such as contributor, free rider, benefit to the group, collective actions). These strategies do not just pick contribution required for entitlement, individual the highest absolute payoffs}they often sacrifice receipt of joint benefit, entitlement to benefit, such payoffs in favor of enduring practices that undercontribution, exploitation, and so on. These generate higher relative payoffs against exploitive evolved conceptual elements are necessary to strategies, when averaged across all interactions. direct the specific motivations that can lead to Antiexploitation computational elements are thus the realization of potential gains in trade while central and indispensable design features of a simultaneously defending against exploitation. cooperative strategy}essential if the strategy is to Properly targeted and calibrated motivational successfully emerge and stably persist. Hence, the circuits are central to defending against free riders relevant question becomes: What exactly do these (and cheaters), and hence essential to making exploitation-resistant strategies look like, when multi-individual cooperation (and dyadic ex- described as psychological (information-proces- change) evolutionarily stable against exploitation. sing) designs? This regulation is delivered through a set of motivational specializations that, for example: The Functional Logic that Regulates Motivation is (1) calibrate willingness to devote effort to an Integral Part of Evolved Social Rationality collective projects, (2) direct punitive sentiment toward free riders, Another cardinal difference between standard and approaches to rationality and evolutionary ap- (3) mobilize reward sentiments towards contri- proaches to rationality involve the role that butors. motivation plays in decision-making. Instead of preferences being arbitrary, wholly acquired, or It is important to emphasize that the control exogenous to rationality, evolved motivational system that regulates the deployment of these specializations are (we believe) intrinsic compo- sentiments is cognitive or computational in nature, nents of more encompassing, ecologically rational and must apply procedures to representations and problem-solving adaptations, such as adaptations inputs to determine, for example, under what con- for 2-party exchange and collective action. That is, ditions to feel punitive sentiment, how much to feel, motivational mechanisms are endogenous and and toward whom to feel it. Despite the frequent indissoluble constituents to systems of rationality mystification of feeling, feelings are not blind, and cognition, with regulatory architectures that ineffable, randomly emerging forces, but are employ evolved, proprietary forms of representa- neurally computed outputs of evolved programs tion in order to direct motivation so that whose structure can be mapped and whose func- individuals correctly implement solutions to adap- tions can be identified (Tooby and Cosmides, 2005).

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 108 J. TOOBY ET AL.

THE EVOLVED FUNCTIONAL DESIGN OF on the other party’s execution of a corresponding THE PSYCHOLOGICAL MACHINERY course of action); and so on. The exchange system UNDERLYING N-PERSON EXCHANGE also involves specialized procedures that interre- late these concepts, map actual situations into Using the foregoing framework, we would like to these representations, evaluate magnitudes, and explore how multi-individual cooperation (n-per- compute necessary regulatory outputs. Two ob- son exchanges, including collective action) as a vious examples are a system that scans the social pervasive human activity can be accounted for. We and instrumental world for opportunities to would like to sketch out some of the cognitive and increase welfare through exchange, and the look- motivational design features that we propose for-cheaters algorithm. underlie the human ability to engage in collective Many of these components are not proprietary enterprises. We would like to link these design to the exchange system, but evolved because they features in a general way to selection pressures and are more broadly useful across a wide array of ancestral situations that we believe led to their functions. For example, to plan an agent needs incorporation into the human psychological archi- mechanisms of valuation}that is, procedures that tecture. can turn representations of a given situation The evolution of the ability to engage in 2-party (actual or imagined) into a magnitude representing exchange would almost certainly have preceded the welfare that the situation holds for the agent. the evolution of the ability to navigate the game Similarly, agents must be equipped with proce- theoretic complexities of n-party exchange and dures that represent potential or actual changes in collective action. Consequently, we would like to situations as positive or negative magnitudes in consider what additional components would have welfare or interest, which generates the perspec- had to be added to cognitive mechanisms for tive-specific concepts of benefits,andcosts. Hu- dyadic social exchange in order to extend the mans can do this for themselves, and attribute this system to n-person exchange, including the special ability to other agents (i.e. have a capacity to case of collective action. Finally, we will consider interpret behavior in terms of an intuitive theory of how such a model fares against criticism that interests that is, we think, one subcomponent of multi-individual cooperation, including collective the evolved system (Baron-Cohen action, could not have evolved via individual et al., 1985)). Interests are representations of the selection for such exchange strategies. sets of changes to situations that positively or negatively modify an agent’s welfare. Acts can simultaneously modify the welfare of more than Dyadic Exchange Builds Many of the Components one party. When this is foreseeable such behavior Necessary for n-Person Exchange is commonly interpreted as expressing a stable The first claim is that the evolution of mechanisms internal variable that regulates how much the for 2-party exchange built a large part of the welfare of the self is traded off against the other computational infrastructure necessary for enga- party: a welfare trade-off ratio or WTR (Tooby ging in n-party exchange (see, e.g. Cosmides and and Cosmides, 2002, 2005, forthcoming). Thus, if Tooby, 1989 for a detailed description of mechan- one party reliably incurs high costs in order to isms; for a review of evidence, see Cosmides and deliver benefits to another, that behavior expresses Tooby, 2005). The evidence indicates that the a high welfare trade-off ratio toward the recipient. social exchange algorithms that evolved for When single exchanges morph into extended exchange use conceptual primitives such as self; series or enduring exchange relationships, they agent (or party); the welfare or interest of a party; may not be structured in the long run by tit-for- exchange; entitlement to benefit, benefit to be tat-like rules (which require alternating acts of gained; cost or requirement to be met to gain reciprocation), but rather by a rule that matches entitlement to the rationed benefit; cheater (a party one exchange partner’s welfare trade-off ratio to that has taken the benefit while having intention- the other’s, scaled by the relative symmetry and ally not met the requirement); intended outcome vs frequency of each party’s opportunities to help the accidental outcome; consent to a social contract (an other (e.g. Tooby and Cosmides, 1996, 2002). This intercontingent plan of action in which each party exchange psychology represents what are, in effect, agrees to undertake a course of action conditional accounts}how much each party has done for the

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 109 other. Accounts and their interrelationships (such does x2 and individual3 does x3 ...’ (or ‘I will if as a ‘balance of trade’ index) generate conceptual you1, you2,you3...will.’). It is important to elements such as owe, debt, obligation, exploitation, recognize that the gain that is produced need not cheating, and so on. Computations on these be a public good, but can be a series of private accounts and their interrelationships also regulate goods (for benefit1, for benefit2, for benefit3...). motivation: Others’ kindness to us moves us (i.e. recalibrates our welfare trade-off ratio toward Discoordination and Complexity as Factors that them so that it is higher than it was before). Their Sharply Limit the Scope of n-Party Exchange indifference, unwillingness to help, or their cheat- ing harden our hearts (i.e. downregulates our One important limiting factor for n-person ex- welfare trade-off ratio with respect to them). If change is the problem of reaching and sustaining exploitive or deceptive enough, these acts may coordination among the interactants. Complexity motivate anger and punitive sentiment towards the in the set of possible alternative arrangements and offender. Obviously, this neurocomputational ma- behavioral interdependencies mounts explosively chinery is beyond conscious awareness, although as the numbers in the exchange grow. The first we are aware of some of its outputs. problem is that of establishing a mutual under- standing among many different minds. In dyadic exchange, this is relatively easy: If the proposed Recursion as One Modification to 2-party Exchange exchange is not worthwhile for one participant, Psychology that can Help to Adapt it to Solving the then the other participant increases the value of Problems Arising from n-party Exchange the offer (or refuses to engage in the exchange). To Understanding the relationships that 2-party ex- persuade a holdout in a multisided negotiation, change embodies seems effortless and intuitive. In however, it is in every other party’s interest to have reality, these interactions involve a surprisingly others contribute the needed additional induce- large number of interrelationships, variables, ment. This leads to an n1 sided game of chicken. steps, conditional branch points, unexpressed Similar negotiative problems and instabilities contingencies, and so on (see Cosmides and emerge, for example, in deciding who is to do Tooby, 1989, for a list of the conditional relation- what, whenever complementary tasks with differ- ships required to encapsulate a simple dyadic ent costs need to be performed. exchange). What makes this maze transparent, Another important limiting factor for n-person intuitive, and navigable is that humans are exchange is the rapidly increasing cognitive equipped with evolved machinery designed to demand posed by the rapid growth of combina- automatically represent and track this intricate torial complexity as the numbers in an exchange set of implicated relationships. What would need grow. For example, if a participant has to divide to be modified in the architecture of 2-party her attention among an increasing number of exchange to allow it to be able to coordinate the exchange partners, how are instances of under- behavior of three, four, or more individuals to contributing or other forms of exploitation to be realize n-person gains in trade? detected? With increasing numbers of participants, Theoretically, much of what is needed could be the problems of negotiation, communication, achieved through a relatively modest change to the coordination, and agreement become far more architecture: adding recursion to the exchange complex. Complexity is injected by the number of structure. The pre-existing computational template different roles individuals could take, by the used for navigating 2-party exchange could be number of different goals the group could pursue, leveraged to navigate n-party exchange through by the possibility of different costs being incurred adding the ability to reduplicate an additional by different participants doing their respective agent-centered unit for representing exchange parts, by the possibility of different individuals relations for each added person. That is, if the 2- deriving different magnitudes of benefit from party form can be abbreviated as ‘I will do x if you the n-sided exchange, and so on. Limitations on do y’, then the generalization of this intercontin- this type of exchange emerge from logistical gent regulatory structure to n-person social ex- constraints in negotiating and implementing change could be abbreviated as ‘For individuals 1 n-sided exchanges, cognitive bottlenecks in though n, I (individual1)willdox1 if individual2 attending and representing, and corresponding

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 110 J. TOOBY ET AL. limitations of the ability to defend oneself against the group in excessively self-interested, group- exploitation within them. These problems are injurious ways, while in followership it encom- relatively small if the set of interactants is small, passes not only general undercontribution, but such as three, four, five, or six. But with increasing also the more specific concepts of unreliability, numbers of interactants, communication becomes disloyalty and disobedience (discoordination). By noisy, mutual surveillance becomes difficult, cheat- effective use of joint attention, persuasion, directed ing becomes easier, and transaction costs become disapproval and approval, threat, punishment, prohibitive. and reward, leaders can collapse discoordinating For these reasons, it is inevitable that in the possibilities entertained by the other n1 members regular course of human life immense numbers of of the group into a single shared representation of mutually beneficial potential gains in trade from n- a plan of action and benefit allocation. The fact person exchange are never realized. Nevertheless, that such a plan is shared itself should excite the human ethnographic record indicates that over greater willingness to engage in an n-person our evolutionary history humans have been able to exchange, precisely because a lack of a shared successfully carve out and master some subsets representation of coordinated action is one of the from the space of n-person exchanges (Ostrom, primary obstacles to achieving n-person gains in 1990; Price et al., 2002). The persistence of these trade. opportunities would have favored selection for Indeed, certain events that invite a common and cognitive and motivational mechanisms that could mutually manifest response can sometimes sub- cost-effectively implement and increase the scope stitute for leadership in solving coordination of the various families of solution. problems. If an event draws joint attention, and inherently invites the same kind of response from most observers, then the necessary coordination to Different Modes of n-Party Exchange Exploit trigger a collective effort can be triggered. The Different Strategies for Solving Coordination right kind of event can spontaneously lead to Problems acephelous defense during attack, hunts precipi- We think that there exist a number of different but tated by the sudden appearance of game , interrelated pathways to solving coordination emergency-provoked rescues, and outrage-pro- problems, and that each has left its imprint on voked mobbing, riots, and so on. The key feature our evolved social psychology. For example, these of event-organized exchanges is existence of a coordinative problems selected for specialized single supersalient event that not only coordinates cognitive and motivational adaptations that create the minds of the co-participants, but makes it clear the uniquely human interlocking roles of leader- for each participant that this coordination is ship and followership. Humans find these concepts present in the minds of the other participants. and behaviors intuitive, and they appear sponta- Implicitly present is a shared appreciation of the neously in all human cultures (within delimited potential for gains offered by coordination, which social contexts). That is, we think a set of activates motivation once the near achievement or neurocomputational adaptations subserving lea- reality of coordination is perceived. dership and followership evolved as solution to the A third pathway that also collapses many of coordination problems that emerge in complex these problems into manageable units is the exchanges and intercontingent social structures activation of a mutually enforced system of strict (Tooby and Cosmides, 2002, forthcoming). Lea- egalitarianism of external conditions and mar- dership allows efficient coordination by routing kers}what Alan Page Fiske calls equality match- coordinative decisions through a single individual ing (Fiske, 1991). This is not an equality of costs, mind so their mutual implications can be compu- benefits, and inner accounts, but of externally tationally integrated rapidly and dynamically. demarcated units of shares and burdens (e.g. one Leadership (when competent) simplifies the dy- man, one vote), greatly simplifying monitoring namics of determining new projects and allocating against exploitation. We think this intuitive efforts to achieve them}something that becomes method of social organization is anchored in our increasingly important when the numbers in a evolved exchange circuitry as one natural mode of potentially cooperating group get larger than a n-person exchange. In general, n-person exchanges handful. Cheating in leadership involves directing are far more cognitively tractable to the extent that

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 111 contributions from the participants consist of short-term end should be low. Correspondingly, parallel behaviors: If everyone is expected to however, the existence of recent and mutually contribute in the same way, then it is vastly easier known precedents gives a coordinative starting to communicate and to monitor such an exchange. point that allows potential interactants to circum- Another natural mode involves collective action vent considering, negotiating, and sorting through to produce public goods, policed by punitive the huge number of logically possible exchange sentiment against free riders (Price et al., 2002; arrangements that arise in situations involving Price, 2003, in press A, B). The entire destabilizing more than a few people. Hence, the prior existence negotiation over the allocation of a jointly of n-person exchange structures should facilitate produced resource or desirable state of affairs is the motivation to engage in such exchanges by eliminated when what is produced is a public offering the opportunity for lower cost coordina- good-like diffused benefit. Collective actions for tion. The magnitude of this effect should be a public goods simplify the issue of defending function of the number of individuals involved. against exploitation by focusing surveillance on The larger the set of minds to be coordinated, the the issue of undercontribution. If the good to be greater the cost of arriving at a new set of mutually produced is not excludable, then producers cannot consistent representations. This means that the be cheated by being denied consumption of the greater the number of participants is, the greater product of the joint effort. As we will explain, the comparative advantage of conservatively many groups that may initially emerge to pursue perpetuating pre-existing arrangements will be, nonpublic goods eventually also produce public however beneficial or flawed those pre-existing goods. This makes the evolved psychology of arrangements were.2 collective action relevant to most and perhaps all The key point is that our evolved coalitional cooperating groups. psychology is designed to appreciate that it is easier to resuscitate or perpetuate successful coordinative arrangements than to create them.3 The Resuscitation or Perpetuation of a Pre-existing In reality, groups do not objectively exist}they n-Person Exchange System is Another Pathway to only exist to the extent that they are represented in Solving the Problems of Coordination mutually consistent ways in the minds of assorted Because successfully coordinated n-person ex- individuals. So, groups only persist if individuals change is so difficult to achieve, its existence once represent them as having continuity over time. Our attained becomes a kind of resource to be valued cognitive circuits for representing groups are and effortfully perpetuated by its participants. advantageously designed to endow them with That is, another condition that greatly simplifies continuity. subsequent coordination and the fruits that it The coordinative arrangements of an existing n- makes available is the prior existence of some person exchange system themselves constitute a previously existing set of understandings that useful resource or instrumentality. This is because arose from a previous coordinated interaction. they make it possible for the individuals in the This means that it is in the interests of participants exchange system (coalition) to consistently take to give the precedents set by a previous n-person advantage of the flow of opportunities to generate coordination a weight and an inertia that they joint benefits that uncoordinated individuals could would not necessarily give those expressed in not. That means that collective efforts do not previous 2-person exchanges, which can be far simply emerge rationally at the time individuals more easily generated, modified, and discarded as come to anticipate the joint benefits that could be needed. The coordination problems inherent in attained from reaching a specific goal, and cease establishing n-person exchanges make it far too when the goal has been reached. Instead, our costly simply to create them de novo as 1-shot minds evolved to inherently value the existence of arrangements in the face of every new opportunity the groups for their own sake, so that we have a for gains in trade. Because the great difficulty of suitable coordinative structure ready and waiting arriving at coordination usually prevents mutually when the need strikes. Just as humans value any beneficial n-person exchanges from coming into other intrinsic good, we are designed to prize the existence where n is large, the default willingness existence of and continuation of the groups we are to invest effort in their creation for a specific members of. We feel pleasure upon becoming a

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 112 J. TOOBY ET AL. valued member of a group, satisfaction in its for a coalitional psychology. Indeed, experimental creation and successes, and sadness at its dissipa- evidence indicates that our minds contain an tion. Our evolved psychology includes a motiva- evolved alliance detector that scans for the tional variable that tracks the social sufficiency of membership in coalitions bound by mutual alli- the group or groups the individual is a member of. ance. This evolved cognitive specialization, as a This variable regulates the appetite for finding or routine part of person representation, takes the developing groups if the individual lacks a pattern of cues of mutual alliance present in sufficient number of sources of support. We seek individual behavior in the social world as input, to acquire, create, and perpetuate group identities. and deduces coalitional categories in the mind as output (Kurzban et al., 2001b). That is, our minds are designed to interpret the social world in terms Groups Tend to Turn into Coalitions, Imposing of coalitions, spontaneously and automatically. Exchange Obligations on their Members Trade-offs arise from the opposing costs and For this reason, there is a tendency for groups to benefits to the individual of belonging to larger outlive whatever rationales, circumstances, or versus smaller coalitions: The larger the coalition, events gave rise to their creation,4 and to gradually the stronger it is, and the more resources and morph into what we call amplification coalitions support it can supply to an individual. However, (Tooby and Cosmides, 2002, forthcoming). That the larger the coalition, the less joint attention and is, the aboriginal n-party exchange system is one value the ‘group mind’ can give to an individual, whose purpose is the amplification of the ability of and the less indispensable she is to it. Even more each of its members to realize her interests in daily important, the larger the coalition, the more events by cost-effectively combining welfare trade- frequently disputes involving welfare trade-offs offs and joint efforts from the other members. will occur among ingroup members rather than Alexandre Dumas memorably encapsulated the between members and nonmembers. Whenever nature of the amplification coalition as all for one disputes are internal to the coalition, the individual and one for all (Dumas, 1844). Dyadic alliances or cannot draw on it for support. At the logical friendships are minimal 2-party instantiations of extreme, if everyone belongs to a single encom- such a coalition, but they can be easily extended to passing coalition, then it is not a coalition that include more individuals. Gangs, cliques, factions, usefully amplifies individual power in the disputes friendships, alliances}these coalitions exist when- that emerge in daily life. Help from the coalition ever individuals feel motivated to provide general cannot be dependably asymmetric when both support to someone else by virtue of implicit or disputants happen to be in the same coalition explicit co-membership. What earns you the right and so qualify for the same support. Reciprocally, to the support of others is the contribution you the smaller the coalition, the more differentially make to supporting others when they could cost- valued and attended to a member will be. More- effectively benefit from assistance. Reciprocally, over, the smaller the coalition, the more indivi- when others observe a pattern of undercontribu- duals lie outside of it, and so more disputes with tion on your part, this erodes the willingness of others will occur between ingroup members and others to support you. outgroup members. As a result, the smaller the For example, disputes are endemic in small-scale coalition, the more often the coalition can be used social interaction, and are often decided in ways to amplify one individual’s power against another that reflect the relative power of the disputants. in a dispute. Consequently, within each larger scale Other things being equal, two individuals have coalition, individuals benefit by calving off a more power than one, three more than two, and so smaller scale coalition to side with them when on, so support for an individual in a dispute by there are internal disputes}support that is earned others is often a considerable help in resolving the by siding with others in other internal disputes. matter in a way that favors the supported (Indeed, more marginal individuals may covertly individual. Everything can be taken from a welcome or encourage disputes among others that powerless individual. So in addition to the other can give them opportunities to assist others and joint resources that were produced by collective hence cement their status as someone entitled to effort ancestrally, ordinary social competition in defense and support.5) As a result, individuals are hunter–gatherer interactions would have selected usually members of many concentric amplification

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 113 coalitions from dyads up to the largest scale of because the well-designed hunter–gatherer was intergroup conflict. Coalitions existing at different unprotected against threats from the social and scales are fractally organized by links of mutual natural world without the support of others alliance. This chronic set of problems suggests that (Tooby and Cosmides, 1996). Individuals have the psychological system that carves out a coali- an evolved appetite for creating coalitional iden- tional identity includes regulatory circuitry to tities independent of and prior to a real need for assess the local fractal structure of groups, and them. If individuals were designed to wait until to allocate effort to forging alliances across a range need struck to build a coalition, the organizational of scales. For similar reasons, each individual in an difficulties and latencies would make it too late. amplification coalition should spontaneously dis- So, to serve these appetites, modern humans inflect approve of the formation of strong individual their lives with notional and recreational groups loyalties toward outgroup members by other that often seem detached from utilitarian need, individuals in the ingroup: Such loyalties will such as churches, clubs, sports teams, political contract the set of circumstances in which the groups, and the like (see, e.g. Tiger, 2004). compromised individual can be expected to Moreover, recurrently spending time with others amplify the power of other ingroup members.6 in group settings is itself a primary generator of the These considerations extend beyond help in expectation that one is in an amplification coali- disputes. If, for example, responsibility for material tion. This means that whatever the formal or legal assistance is diffused among too many individuals, theory present, life in the workplace invites no one helps (leading to the well-known bystander implicit assumptions of mutual amplification and effect; Darley and Latane, 1968). To be cheated in support that transcends work-specific activities. this context is to fail to receive assistance when one Conformity to these expectations is considered has given it to other ingroup members. Cheating elementary decency, and their repudiation or among large undifferentiated groups is more likely discouragement is experienced as an alienating because the blame for the absence of assistance is feature of many corporate and nonprofit work- distributed among too many others. For assurance place cultures. of help from others to be strong, there must be bonds of mutual dependence among a small enough The Status of Coalitions is a Public Good, Making number of individuals that withdrawal by a cheated Coalitions Collective Actions individual strongly hurts a few specific others. Individuals accordingly feel a sense of precarious- Amplification coalitions are intuitively under- ness among homogeneity, and a desire to be stood, spontaneously self-organizing, and may individually valued by a small number of local emerge into awareness when one or more indivi- others (see also, Tooby and Cosmides, 1996). duals are challenged by outsiders. One crucible Indeed, Marilyn Brewer proposed that individual that led to the evolution of the psychology identification with groups is shaped by a motivation underlying amplification coalitions was their an- for group optimal distinctiveness, which may also cestral operation in zero-sum competitions over serve this function (Brewer, 1991). In short, it is not access to resources by contending sets of indivi- enough just to be nominally a member of a duals. Individuals take or relinquish resources to coalition}a sufficient number of local individuals the extent that they feel empowered or over- in the coalition must know you, differentially value matched. Public signals of support (or its absence) you, monitor your welfare, and support you when lead individuals and sets of individuals to revise you need it. what they attempt to possess, consume, or do. In Both for reasons of social competition and consequence, our species-typical psychology material insurance, when homogeneous coalitions evolved to represent coalitions as having status get too large, our evolved coalitional psychologies (Brase, 1998; Sidanius and Pratto, 1999). Indivi- prefer to give them internal factional structure. As duals (where cost-effective) expend effort to Madison pointed out in the Federalist Papers, the defend, advertise or enhance the status of the ‘latent causes of faction are thus sown in the coalitions they belong to, and to deflate the status nature of man’ (Madison, 1788/1982). More of rival groups. This is one set of individual broadly, modern humans feel naked without the behaviors that humans are designed to interpret as protection of a sufficient set of group identities expressions of the group as an entity. For a given

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 114 J. TOOBY ET AL. action by individuals, this interpretive system of the group (fully paid up; not a free rider)? If the licenses the transformation of He did this to me group has been in existence for some time, new (one individual taking action with respect to members at the threshold of joining will intrinsi- another) into They did this to us (one group taking cally strike existing members as resembling chea- action with respect to another). ters in that they are attempting to gain the benefits A central feature of interactions over group of membership, without having yet paid any costs status is that their product is a public good. That (Tooby and Cosmides, 2002). Social practices will is, anyone who is a member of an amplification emerge to express or disarm the punitive sentiment coalition automatically partakes of its status, and as well as the distrust recruits spontaneously gets the benefits in daily transactions of the greater attract from veterans. A strong feature of coali- weight nonmembers will place on members’ tional psychology will be organized around issues interests out of respect for how members coordi- of membership, group identity, the price of entry, nate to support each other in conflicts. Acts of initiation, genuineness of membership (loyalty, alliance cause observers to categorize individuals commitment), exclusivity, and the pressure on together as members of the same coalition new members to make contributions to bring them (Kurzban et al., 2001b). The representations in into the implicit exchange relationship in the the minds of observers of the status of a coalition minds of veteran members. This exchange view are a common resource that has the properties predicts that organizations in all human cultures diagnostic of a public good. That makes the work will spontaneously tend to pass costlier, less of maintaining group status intrinsically a collec- attractive tasks to younger, newer recruits, and tive action (Tooby and Cosmides, 2002, forth- recruits will be motivated to sacrifice more in order coming). Because amplification coalitions arise to make more dramatic and observable contribu- easily and spontaneously from the mere existence tions than established members will. The punitive of any cooperative precedent, and individuals tend sentiment veterans feel towards new entrants, to assume their existence and continuity, it follows combined with the advantages for veterans of that all groups will gravitate toward being creating strong representations of the dominance experienced, to some extent, as collective actions of the coalition over new members, together often by their members. This will be especially true lead to the coordinated infliction of costs by around issues that affect the status of the group in veterans on new entrants. Correspondingly, when the mind of outsiders. recent or potential entrants assume increasing Amplification coalitions by their nature are costs attendant upon membership, the motiva- assumed to continue indefinitely as ongoing tional systems of established members will increas- operations, so they are formally different from ingly recategorize them as potentially worthwhile collective actions that exist to achieve a realizable, recipients of group benefits in an n-person finite goal, with a specific termination point. exchange. The motivation to police the boundary Extended cooperative interactions invite the im- will be calibrated by the magnitude of the benefits plicit assumption of the existence of an amplifica- that go with membership, the labor recruitment tion coalition, binding its members into an needs of the coalition, the levels of trust and exchange relationship. The individual is never interdependence needed to function as a coalition, done with the obligation to contribute to the and the frequency with which the welfare of the group so long as the group persists. One can cheat coalition (or its participants) depends upon sub- by not contributing one’s share of support to stantial sacrifices by members. others in the coalition, including not contributing one’s share to maintaining the representation of The Addition of the Cognitive Ability to Represent group status in the minds of nonmembers. a Group as an Individual Allows an Additional Individuals can also cheat by assuming member- Approach to Applying Dyadic Exchange Machinery ship and signaling it to outsiders without having to n-person Exchange earned it (Brase, 1998). This makes policing the boundaries of such a coalition a sensitive issue. Although the addition of recursion is one im- Natural questions that express this intuitive portant computational modification to 2-party coalitional psychology include: Is individual i really exchange psychology, a second important mod- a member of the group? Is i a contributing member ification to this exchange psychology involves

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 115 widening the exchange system’s definition of what first taking the individual’s perspective, and then kinds of entities can be computationally treated as the group’s (represented as a single party). This a party or agent. During the initial evolution of allows dyadic exchange psychology to generate exchange circuitry, party would have meant and cognitively treat many kinds of social rules as individual human. The next step would have been social contracts (Cosmides and Tooby, 1989). widening party or agent so that open arguments in The widening of the agent slot to include groups procedures that formerly would have referred as well as individuals is a cognitive expansion that strictly to human individuals become able to refer applies beyond exchange relationships, although also to coalitions, alliances, communities, and that may have been its selective crucible. In human other entities as well. This allows humans to psychology, groups are not treated simply as represent, understand, engage in, enforce, and categories or aggregations of individuals. Groups cheat on dyadic exchanges where one party is an are conceptualized and experienced as entities to individual, and the other party is a group, as well which we can properly attribute mental states, and as where both parties are groups. which can have propositional attitudes (Cosmides This cognitive ability to interpret a coalition as and Tooby, 2000). That is, humans have no an agent is not just speculation: Two decades of trouble interpreting claims about the mental states research on social exchange reasoning demon- of groups as if a group constituted an individual strates that subjects can understand and reason agent with a single mind. (In reality, obviously, about 2-party exchanges just as readily when the they are not.) For example: Our university or exchange rule is a social rule entered into with a department or company can want us to do things; group as they can about exchanges between two our community can be angry, or disgusted or individuals (Cosmides, 1989; Cosmides and Too- pleased with us; our nation can be intimidated or by, 1992, 2005). Such a broadening of the meaning vindictive; our village can mistakenly believe we of agent creates alternative ways for individuals to killed someone; we can be afraid to disappoint our represent and enter into n-person exchanges. An group. Nelson signaled at Trafalgar, ‘England interaction could be represented as a single multi- expects that every man will do his duty’ without lateral n-person exchange with every individual as wondering whether his fleet would be philosophi- a separate node; or, it could be represented for cally puzzled about how a nation can entertain an each individual as a dyadic exchange between the expectation. individual and the group, with two nodes for the Moreover, humans also unproblematically treat two agents}self and group. Mathematically, an n- groups according to an intuitive theory of interests person multilateral exchange involves nðn 1Þ that we believe originally evolved to interpret directional links, while an n-person exchange individual action. Humans not only represent conceptualized as a series of individual to group individuals as having interests, but typically exchanges involves 2n links (or n exchanges). represent groups has having welfares or interests Seeing an n-person exchange in terms of a series as well. Accordingly, groups (like individuals) are of dyadic exchanges between each group member experienced as having a unitary exchange account and the group as a single entity allows for great that we can trade off our welfare against (and vice cognitive simplification and transparency (when n versa). In so doing, we can owe the group, supply is greater than 3): Simply using pre-existing dyadic help to the group so that it becomes indebted to us, machinery, each individual can clearly understand feel entitled to group assistance, or punish the and reason about what she herself owes the group, group for not taking our welfare sufficiently into and what the group owes her. She does not need to account; reciprocally, the group can owe us, simultaneously represent links to every individual supply help to us so that we become indebted to in the group. She can also monitor cheating on the it; it can feel entitled to our assistance, or punish us part of other members applying dyadic reasoning for not taking its welfare sufficiently into account. parallel to what the individual applies to the self: Is Groups can have status, rank, stigma, and a given individual taking the benefit of the joint dominance relations, not to mention friendships effort? Is the individual holding up their part? A and enmities. We can see them as exploitive person can comprehend and cheater detect an n- (favoring their own welfare above ours) or person exchange as a series of dyadic exchanges by generous (injuring their own welfare to benefit going individual by individual through the group, us). A group can be dominant over us, whether we

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 116 J. TOOBY ET AL. are a member or not. (Indeed, leading members of How humans represent the organizations they male fraternal organizations often want new work within will have a major impact on how they members to appreciate the power the coalition behave in these organizations. Consequently, it is has over them; Tiger, 2004). Many cognitive hard to overestimate the magnitude of the impact scientists are persuaded by the developmental, that the ability to represent groups as individuals cognitive, and neural evidence that humans come has had on human sociality. This key expansion in equipped with what has been called a theory of the human ability to engage in n-person exchange mind, a specialized set of mechanisms that inter- populates our mental world and our social world prets individual behavior in terms of mental states with groups, using much of the same machinery (Baron-Cohen et al., 1985). We propose that a set that governs how we think and feel about of augmentations and modifications to the theory individuals. Yet, we maintain in parallel the ability of mind system constitute a theory of group mind to interpret groups as sets of individuals engaged system (Tooby and Cosmides, 2002, forthcoming). in multilateral exchange. These alternative ways of This system generates representations about what representing n-person exchanges are different, and a group thinks and feels, triggers the individual’s mobilize different ways of thinking and feeling: In feelings with respect to imagined mental states of one case, I am in a multilateral exchange with groups, and treats groups as entities that can be everyone else in the group. In the other, I am in a the object of emotions and motivations that dyadic exchange relationship between myself and initially evolved for individuals. the group. It is possible for the same individual to In reality, of course, groups (as sets of indivi- activate both of these representational modalities duals) do not have a single mind, body, or voice. toward the same object, at least at different times. In order to treat groups as agents, the mind must A corporation might for some purposes be contain procedures that can interpret the obser- represented by the employee as a single agent vable acts of individuals as the behavior of groups (the company) in a dyadic exchange relationship or the expression of group mental states and with the employee, and on other occasions as a intentions. If groups are mentally represented as team or amplification coalition (a multilateral amplification coalitions, then this predicts that one exchange) that the employee is a member of. major set of acts of individuals that invite These alternative representational methods will reinterpretation as acts of the group are behaviors activate different motivational responses. by one or more individuals in one coalition that It is perhaps not an overstatement to say that negatively impact the welfare of one or more corporations (and many other formal organiza- individuals in another coalition. When members of tions) as human social institutions are founded on the Los Angeles African–American community the evolved representational ability to represent a saw footage of several Los Angeles police officers group (i.e. a corporation) as a unified agent or clubbing Rodney King, they did not treat it as an person, while collectives, associations, and unions affair just between the interactants}it was seen as more often invite representation as multilateral one community (coalitional entity) doing some- exchanges among (relatively) equal members. If I thing to another community (coalitional entity). In interpret my employment with a corporation as accepting or attacking the acts of individuals from dyadic relationship, then my attention is focused other coalitions, coalitions act as if they are on allocational conflict over benefits and costs (e.g. negotiating general precedents specifying how salary, effort, risk) between the two parties, myself much members of each coalition are obliged to and the corporation. Ordinarily, in dyadic ex- trade off their welfare on behalf of members of the change, the marginal cost of moving a unit of other coalition (an equilibrium welfare trade-off benefit from one individual to the other is ratio expressing relative power or dominance). compared, with some modulated expectation that Some individual acts}outrages}serve as coordi- units will go to the individual to whom it does the native signals for a group to act together in a most good. If the corporation is viewed as an collective action to obtain what is usually a public individual, however, it strikes the employee as a good: to act together to change the representations wealthy, dominant, and potentially exploitive in the minds of the other group as to the relative individual}something not significantly harmed power of the two groups, by threatening, attack- by transferring marginal units of benefit that ing, or dispossessing them. would make a great deal of difference to the

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 117 employee. The fact that the corporation does not should the outputs of this motivational component respond according to this logic is a continuous look like? source of alienation for employees. This modality The problem of sustaining cooperation in the of representation provides the foundation for the face of potential exploitation is severe and Marxian and labor union worldviews that frame ubiquitous: After all, the ordinary psychology of workers and management as inherently locked in effort minimization should make everyone a an oppositional, zero sum relationship with each tactical free rider. In all conditions, including other. Alternatively, if the individuals networked exchange, motivational adaptations in humans together into a corporation represent themselves should have been selected to identify and curtail as members of a team, then the invited motiva- unnecessary investment (i.e. units of effort whose tional consequences are a convergence of interests. investment does not trigger or sustain returns). In either case, workplaces usually bring together Hence, our evolved cognitive equipment should bounded sets of individuals in continuous associa- connect the motivation to expend effort in an tion with each other, often facing an open-ended exchange to an assessment of the sensitivity of set of situations. Ancestrally, spending a major others’ responses to one’s own contributions, both fraction of each day together indicated mutual upwards and downwards. That is, do they increase membership in an amplification coalition, impli- their contributions when I increase mine? Do they citly obliging members to provide a form of downregulate their contributions when I down- cooperative social insurance to each other. The regulate mine? Are they appreciative if I sacrifice expectations generated by different forms of for the joint effort? Do they even notice if I do not? organization are not just provided by explicit rules As a result, the motivational component should or by local cultural practices. Our evolved have been designed by evolution to modulate one’s mechanisms, triggered by experience, assign their own level of compliance to one’s obligations in an own system of evolved meanings to social interac- exchange in response to others’ levels of compli- tions even in defiance of explicit rules or attempts ance (or at least to others’ monitoring and at contrary cultural indoctrination. eventual responsiveness). In dyadic exchange, I am no longer motivated to live up to my part of the exchange when my partner is not living up to hers. In n-person exchange, motivational levels Anti-exploitation Motivational Circuitry in should be a function of the distribution of n-person Exchange contributions by others. In general, such circuitry Neither dyadic nor n-person exchange can evolve motivates greater effort when others are investing or be evolutionarily stable unless cheating is sufficiently (and contingently), and curtailed effort detected and then effectively responded to. Nu- in the presence of free riding (or indiscriminate merous experiments have established the first investing). This set of conditional rules regulating element: that the human brain does indeed contain the motivation to participate is an important evolved circuitry specialized for detecting cheating first line of defense against being exploited by in both dyadic and n-person exchange relation- others (although it permits the conditional exploi- ships (Cosmides, 1989; Cosmides and Tooby, tation of others). 1989, 1992). To function properly, however, Accordingly, our evolved motivational system cheater detection circuitry needs to be coupled to should be designed to recognize situations where a component that motivates effective responses to others systematically benefit from one’s own cheating. An effective response is one that helps to sacrifices without contributing proportionately. make participation in exchange fitter than either Such situations should be cognitively represented cheating or nonparticipation (as well as fitter than as distinct from other situations, and marked as other, more complex uncooperative strategies). intrinsically motivationally significant. The detec- The key property that these motivational circuits tion of such a situation (or its possibility) should must have is the effect of making the average then trigger motivational outputs designed to lifetime fitness of those equipped with cooperative effectively redress the potentially adverse fitness circuits, on balance, greater than the average ordering manifested in an exploitive situation (i.e. lifetime fitness of those with a stronger disposition some property of the cooperative strategy must to free ride. Given this design criterion, what insure that on average, noncooperative exploitive

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 118 J. TOOBY ET AL. strategies have lower fitness, at least when they relative payoff should (under many conditions) be become common). Hence, procedures for social selected to be perceived as worse than a zero comparison of effort and benefit should be very absolute payoff that is equally bad for both important motivationally. Of course, the under- participants. Widespread evidence from experi- lying logic of exchange involves delivering benefits mental economics suggests that this is the case, only when the cost of providing them is less than such as the high level of rejection of offers in the the benefit produced}no one expects you to cut ultimatum game (Hoffman et al., 1998; Henrich off your arm to make dinner for a friend. Given and Smith, 2004). Indeed, the sucker’s payoff (in the wide range of cooperative endeavors humans prisoner’s dilemma situations) should not just be engage in, this means that cooperators will usually aversive because it is low paying in an absolute need to represent not just behavior but also the sense, but because it is relatively low paying underlying costs and benefits that are associated compared to the gains of the other interactant. with behaviors. If your child is starving, I do not So, algorithms in the brain characterize situations expect you to give me food, even if I previously fed in terms of whether others are benefiting from the you. We operate in a world where the inferred individual’s own sacrifices in a way that is costs and benefits of actions are relevant to our disproportionate to their contributions. For con- understanding of our exchange relations. venience, we will call this an evolved characteriza- Accordingly, to engage in n-party exchange tion of the degree of personal fairness or personal behavior, the system needs to represent the exploitation exhibited by a situation. As history accounts own welfare and group welfare in order and evolutionary theory both attest, evolved to make decisions governing one’s own allocations mechanisms in the mind do not have an automatic between these two accounts. To do this, the mind appetite for global fairness of all to all (including needs to compute a regulatory variable, the welfare sacrificing personal gains to be fair to others). Our trade-off ratio (WTR), that governs what trade- evolved mechanisms are not designed to intrinsi- offs (sacrifices) the individual will be willing to cally object to one’s own unfairness to third parties engage in that have a negative impact on own (although in the right social environment such welfare and a positive effect on group welfare (and attitudes may emerge). Our evolved exchange vice versa). To protect itself against exploitation, psychology is designed instead to manifest a the system also needs to be able to monitor and distaste for exploitation or unfairness by others represent events and actions in terms of their to oneself, as well as by others to family members, significance with respect to others’ welfare ac- to a friend or ally, to an ingroup member, or counts (e.g. the contributions others make, the (through putting a group into the agent slot) to costs they incur in making contributions, the personally valued groups (see, Fehr and Fischba- benefits they derive from the exchange, the impacts cher, 2003, for a different view). I have on them, etc.). These variables need to be There are several distinct motivational outputs tracked both for individuals, and (for some from this system, when it recognizes a situation of purposes) on a pooled basis. This enables the personal exploitation, in addition to contribution motivational system to be able to compare one’s downregulation. The balance of costs and benefits own welfare trade-off ratio to the welfare trade-off arising from the situation the individual is in ratio that others are using, and therefore to determine which motivational output and beha- determine whether the actor is being exploited by vioral response will dominate. Most fundamen- others (or is exploiting others). Our motivational tally, the exploitive aspect of a situation should be systems switch into different modes of activation experienced and represented as intrinsically un- depending on the answer to questions like: Are pleasant or aversive, independent of the other others contributing less (more) than I am? (e.g. Price payoffs of the situation, so that the prospect of et al., 2002; Price, in press B). removing exploitation presents itself as a goal (if The evolved aversiveness of low payoffs in achievable) with a positive payoff. Not only should interactions should be a function of comparative exploitation be aversive, but individuals who payoff, not just absolute payoff. In rational choice consistently exploit more than others should theory, any payoff is better than no payoff, but in themselves come to be seen as aversive. Other the evolutionary competition of alternative de- motivated responses that defend against adverse signs, a positive absolute payoff that is a low selection favoring free riders include: punishing

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 119 free riders (see, e.g. Price et al., 2002); avoiding participant, others must match it, or their failure individuals who differentially undercontribute; to do so will establish the future collective ceiling leaving situations in which one is being exploited; below the optimum. refusing to initiate exchanges in which exploitation So, our evolved exchange psychology contains is likely to occur; refusing to continue in exchanges procedures that motivate participation in joint in which exploitation is occurring; aggressively efforts when they can be detected and assessed as forcing those disposed to free riding to contribute; beneficial. Defense against exploitation requires and driving off those disposed to free riding. It is monitoring of others’ actions and their contingen- unparsimonious to assume that all of these cies. We believe that different structures of responses were independently selected for. Rather, monitoring have selected for different modes of it seems plausible that many or all these responses n-party exchange. These include joint monitoring are generated by a single underlying motivational (where everyone in a collective enterprise monitors element: finding personal exploitation intrinsically everyone else), leadership monitoring (where single aversive. individuals}‘leaders’}monitor and direct enfor- cement), asymmetric monitoring (where multiple individuals with disproportionate stakes in the Opening Cooperative Moves and Subsequent joint effort or lower costs of enforcement regulate Monitoring Forms in n-person Exchange the effort), role monitoring (where specialists Because mutual benefit through coordinated cultivate a socially valued identity based on n-party exchange is hard to achieve and easy to monitoring), or some mixture of them. Joint undermine, joint contribution levels will often be monitoring was perhaps the first to evolve, but far lower than optimum. The recurrent opportu- depends on the practical opportunity for small sets nity to capture these underrealized benefits of individuals to directly observe each other’s selected for design elements that promote upward levels of participation. All of these organizational movement. Perhaps the most important of these alternatives are anchored in an evolved psychology included an initial cooperative orientation parallel of social comparison that evolved in the context of to tit-for-tat’s opening cooperative move (Tooby n-party exchange. and Cosmides, 2000). This orientation involves a Indeed, a standard finding of experimental readiness to make the first move (at least where the collective actions (public good games), is that the number of parties is not too large). This poten- majority of subjects are conditional cooperators, tially initiates new upward movements at the i.e. they cooperate more when they perceive that beginning of any new event boundary that co-players are more willing to cooperate (Orbell plausibly invites new collective projects, coupled and Dawes, 1991, 1993; Ledyard, 1995; Fischba- to a decision-rule to modulate downwards if the cher et al., 2001; Lubell and Scholz, 2001), and less effort goes unmatched. Additionally, we recognize when they believe that co-players are free riding and respond to others’ supernormative initiatives (Fehr and Ga¨chter, 2000; Kurzban et al., 2001a). (we can be ‘inspired’ by others, a complement to Moreover, a standard finding of anonymous our being cooled by others’ hypocrisy or defec- public good games, in which players have no tion). We are proud of our own supernormative information about co-player cooperativeness at efforts, especially if they successfully invite others the start of the game, is that cooperative behavior to new levels of commitment, and shamed if our follows the pattern predicted from an evolutionary efforts are discovered to be lower than others, and perspective. Contributing is highest at the outset of to have inhibited positive interactions. As a the game (players open with invitations to general rule, the motivational system should be cooperate), and then decays gradually over time, more willing to make an investment if it is public; as players receive information that some co- investments should preferentially be made in a players are free riding (Ledyard, 1995; Masclet continuous flow of consecutive increments (where et al., 2003). This gradual decay apparently occurs this is not inconsistent with public delivery), so because higher contributors in initial rounds that their magnitude can be modulated contingent ratchet down their contributions as the game on others’ degree of matching}a pattern that progresses, in order to avoid exploitation by reduces the opportunity for free riding. Once matching the average expected co-player contribu- a higher contribution level is made by one tion. The average constantly dwindles due to

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 120 J. TOOBY ET AL. persistent free riding (Fischbacher et al., 2001; termination of an otherwise mutually beneficial Kurzban et al., 2001a,b). ongoing intercontingent effort. Consequently, un- Further, as expected if cooperators are sifting less maintained by corrective social feedback, for exploitation, interactants appear to monitor cheating or undercontribution will spread conta- one another not just constantly, but accurately. giously, and contributions to joint efforts will Among villagers in a hunter–horticultural society, ratchet downwards (Kurzban et al., 2001a). When perceptions of co-villagers’ engagement in general opportunities for monitoring and feedback are pro-village altruism correlated positively with inadequate, effort minimization tends to operate measures of co-villagers’ actual engagement in unopposed, and the cooperative network erodes specific altruistic activities (Price, 2003). In a away. sugarcane cultivating workgroup in this same The key point is that free riding is not just an society, perceptions of co-worker cooperativeness inefficiency, an injustice, or a marginal strategy. (attendance record and physical effort in work How free riding is treated is the central determi- sessions) correlated positively with more objective nant of the survival and health of cooperative measures of this cooperativeness (Price, in press organizations. Given the existence of motivational A). Moreover, participants accurately distin- circuitry that evolved to defend against exploita- guished ‘intentional’ low contributors (those who tion, social penetration by free riders is a brake or could have contributed highly but chose not to) even an off switch, turning cooperation down or from ‘unintentional’ low contributors (those un- off among individuals whose motivation to co- able to contribute highly). operate is dynamically interlinked. The social cost of free riding is not just the marginal effort that A psychology of conditional punishment co-evolved free riders do not contribute nor is it primarily the with a psychology of n-person exchange. The benefits they consume but do not earn: The evolved motivational mechanisms regulating n- greatest cost that free riders inflict is the loss of person exchange are designed to make each all the potential gains from n-party exchanges that individual’s actions contingent on the actions of otherwise would have been achieved if free riding the others in the cooperative interaction. Conse- had not triggered antiexploitation motivational quently, these mechanisms endow joint efforts with defenses among cooperators. That is, by forestalling dynamical properties. That is, an n-person ex- the emergence or continuation of a range of change is a sensitive network of intercontingent mutually beneficial n-person exchanges that would feedback loops. Depending on variables such as otherwise exist, free riders can and do inflict huge the opportunities for free riding and the frequency costs on cooperators out of all proportion to the of free riders, an intercontingent cooperative parasitic benefits they derive. structure can dynamically drive itself toward One can approach the same conclusion by higher levels of cooperation or downwards toward considering the likely sequence of steps in the lower levels of cooperation or dissolution (Ehrhart evolution of the psychology of cooperation: In and Keser, 1999; Fehr and Gachter, 2000; Masclet general, any strategy for exchange that reliably et al., 2003). For example, if many participants are allows cheaters to outcompete cooperators cannot committed and enthusiastic, then the probability evolutionarily persist. But (under reasonable con- that others will join and contribute is increased. In ditions) cheaters cannot outcompete those coop- contrast, the presence of undercontributors (chea- erators designed to follow a strict conditional ters, free riders) in the network inhibits contribu- exchange strategy involving the detection of tions by others. Individuals are more reluctant to cheaters and the refusal to enter into exchanges join collective efforts in which they can anticipate with them. Plausibly, the establishment of this that others will be benefiting without contributing. first-order exchange strategy as part of our Indeed, the presence of individuals with a track ancestral species-typical design was thus an early record of greater free riding should prevent many phase in the evolution of our psychology of n-person exchanges from coming into existence in cooperation (Cosmides and Tooby, 1989). More- the first place. At a minimum, the presence of free over, this conditional strategy is reasonably riders degrades the performance of the network at efficient for dyadic exchange. Individuals who realizing potential joint gains. More seriously, abandon or refuse dyadic exchanges with cheaters sufficient penetration by free riding can trigger the are only forgoing interactions in which they are

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 121 unlikely to get benefits anyway. However, as riders stand in the way of reaching valuable gains n increases for n-party exchanges, abandonment through cooperation. The first line of defense for of the exchange in the presence of a cheater conditional cooperators}abandoning the ex- becomes increasingly costly and inefficient in terms change in the presence of free riders}is very costly of forgone benefits. Under strictly conditional to conditional cooperators. As a result, the presence cooperation, no matter how many cooperators of free riders in cooperative contexts was ancestrally are present, a set of interacting cooperators can be (asitisnow)aseriousadaptiveproblemfacing prevented from beneficial interactions by the conditional cooperators. If they exist, selection presence of a single cheater. Naturally, the greater would have strongly favored the subsequent evolu- the number of individuals involved, the more likely tionary emergence of psychological mechanisms that one (or more) will be a cheater, preventing the (1) can realize gains from n-person exchanges even in exchange (assuming an early ancestral social the presence of cheaters, while simultaneously (2) in which individuals are equipped only preventing cheaters from outcompeting cooperators. with a psychology of strict exchange). Such a Are there any such designs? strategy insures that only exchanges with very An evolutionarily tailored motivational system small numbers of individuals will ever take place, directing punitive sentiment can help to solve the leaving the rest as a large, unachieved residual adaptive problem posed by free riders (Price et al., class. Of course, despite its inability to capture 2002). If the presence of a small number of free gains from larger scale exchanges, a strategy of riders (or one) is inhibiting optimal contribution strictly conditional exchange would nevertheless levels to an n-person exchange, then punishment have been strongly favored by selection because it directed at the individual will drive her off, or did allow capturing gains from smaller scale induce her to contribute. In either case, the exchanges. Indeed, for our hunter–gatherer ances- situation is no longer psychologically marked for tors, a substantial number of exchanges were very cooperators by personal exploitation, and so they small-scale}dyads, triads, tetrads, and so on. are no longer inhibited from contributing effec- Even now, in modern mass society, the great tively to the exchange. Equally, in selectionist majority of daily cooperative interactions involve terms, free riders are no longer outcompeting only a handful of individuals. contributors. That is, it is the dynamic sensitivity In short, a strictly conditional exchange strategy of n-person exchanges to the presence of exploita- would have evolutionarily emerged and been pro- tion that selects for punishment. Rather than pelled upward toward species-typicality among our inefficiently removing the exploitation by depriv- ancestors because (1) it cannot be outcompeted by ing a number of potential cooperators of the an exploitive strategy, (2) it is able to take advantage benefits of n-person exchange, punishment can be of any n-person exchange not involving a free rider, efficiently targeted exactly at the source of the and (3) opportunities for such exchanges were problem, the exploitive individuals themselves. ancestrally ubiquitous. This makes this strategy fitter Hence, in an ancestral social environment of than strategies that do not engage in exchanges, or potential n-person exchanges, the evolution of that free ride (cheat). We think that these selection mechanisms directing punitive sentiments toward pressures established something resembling a strictly free riders was favored by natural selection. conditional psychology of exchange among our Because punitive acts and the evolved dispositions ancestors as a platform out of which a more that cause them unlocked access to a wide range of complex psychology of cooperation subsequently previously unattained cooperative gains, the cir- evolved. Accordingly, we should not expect to see a cuitry underlying this punitive psychology was first-order cooperative psychology whose motivated advantageous. These fitness advantages increased contributions are unaffected by the presence of free its frequency until it became (we believe) a riders, because such mutant designs would have been significant part of our species-typical social psy- exploited and outcompeted. What we predict on chology of cooperation. Widespread experimental selectionist grounds is what we observe empirically: evidence indicates both that humans do have The presence of free riders downregulates coopera- punitive sentiments toward exploiters in n-person tive behavior. exchanges, and that n-person exchanges can be Because the detection of free riders by coopera- driven to higher levels of cooperation when tors inhibits contribution to joint efforts, free subjects have the ability to punish (Fehr and

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 122 J. TOOBY ET AL.

Gachter, 2000). Although the strategy of avoiding will be required to comprehensively rule out the cheaters can work as the sole defense against hypothesis that punitive sentiment was secondarily exploitation when exchange interactions are small designed by selection to serve as a system for (dyads, triads, tetrads, etc.), the larger the size of recruiting additional labor into n-person the interaction, the more efficient punitive co- exchanges. operation strategies become at sustaining coopera- What benefits do punitive strategists derive from tive organization. their evolved motivational design and the expen- ditures of effort it causes them to make? Why The design features and evolutionary dynamics of a would a cooperator equipped with punitive psychology of punitively defended conditional co- motivation toward exploiters be fitter than co- operation. For the reasons outlined above, we operators without punitive motivation? Punitive think that the human mind contains an evolved, strategists switch on the productive possibilities of functionally specialized motivational mechanism the groups they are in, unleashing collective efforts that, when exposed to a situation of personal that would otherwise be inhibited by the presence exploitation, generates a punitive sentiment to- of free riders. This happens because the presence ward the agent that is deriving an unfair advantage of punitive strategists in potential exchange inter- in an exchange. We think evidence shows that this actions repels free riders, causing them either to mechanism becomes strongly activated in collec- avoid such interactions or to become (faculta- tive actions, and evolved as an anti-free rider tively, in the presence of punitive strategists) device (Price et al., 2002, in preparation; Price, in behavioral cooperators. Because free riders avoid press B; Tooby and Cosmides, 2002, forthcoming). punitive strategists, punitive strategists will far In n-person exchanges, this sentiment motivates more often find themselves in groups without free individuals to inflict a cost on free riders (ex- riders. Punitive strategists, unlike those who are ploiters) sufficient to reorder the net benefits that simply strictly conditional cooperators, do not the interactants derive from the exchange (when have to forgo large numbers of opportunities to this can be done cost-effectively). That is, in the harvest a valuable joint gain. Even more impor- absence of corrective action, free riders are better tant, the advantage of a punitive strategy applies off than cooperators. After corrective (punitive) at small scales even when it is rare. For example, it action, free riders are made worse off than will be less often cheated even in dyads, because contributors. The functional product of punish- the cheating triggers punishment, and so the ment is an outcome in which the most exploitive temptation to defect is lowered by the anticipation individuals either end up with a lower welfare than of punishment7. Punitive strategists can even contributors, or end up less well off than they afford to be more trusting, and can more success- would have been if they had not attempted to fully enter into one-shot dyadic exchanges (pro- derive benefits from the collective effort. The vided their interactants can anticipate that their presence of punitive strategists (punitive coopera- partners will have a subsequent opportunity to tors) changes the payoff structure for free riders, punish). In sum, anti-free rider punitive strategies influencing their choices (to the extent they unlock access to a wide range of n-person respond to anticipated payoffs). When disposi- exchange interactions that would otherwise have tional free riders (individuals psychologically gone unrealized. designed to not contribute) detect that one or A number of scholars, such as Boyd and more punitive individuals are to be in an n-person Richerson, argue that punishment emerged exchange interaction, they should avoid the inter- through cultural or biological group selection or action}which will be low paying and hence as a product of gene–culture coevolutionary aversive for them. When facultative (tactical) dynamics}i.e. that punishment of norm violation free riders detect interactions involving punitive is altruistic in the biological sense (Boyd and strategists, they should avoid the interaction Richerson, 1992; Boyd et al., 2003; Henrich, 2004). or should tactically contribute. At present, the The central argument is that mutant punitive weight of evidence appears to support the view cooperators and nonpunitive cooperators would that the primary function of punitive sentiment is share the same group-wide benefits of punishment, as an anti-exploitation defense (Price et al., 2002; but that only the punitive would bear the costs Price, in press A, B). Nevertheless, more work of inflicting punishment. So, on this view,

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 123 nonpunitive cooperators would outreproduce pu- of individuals, then mutant punitive strategists nitive strategists in the same group in the same way would indeed have a hard time outcompeting that free riders displace indiscriminate coopera- nonpunitive cooperators, because both would get tors. The punishment of free riders is viewed as a the benefits while only the punitive would pay the collective action problem (Yamagishi, 1986), and costs. But because ordinary sociality consists of the evolution of punishment is considered to suffer numerous daily interactions composed of dyads, from a second-order free rider problem. Therefore, triads, and so on, nonpunitive individuals will according to this line of thinking, motivational commonly find themselves in potential exchange mechanisms designed to punish free riders cannot interactions without the punitive individuals, even evolve without group selection or gene–culture when the relative frequency of punitive individuals coevolutionary dynamics (Boyd and Richerson is high. For example, whenever a nonpunitive 1992; Boyd et al., 2003; Henrich, 2004; see also cooperator is in a dyad with a free rider, they will Panchanathan and Boyd, 2004). not be in a dyad with a punitive strategist. The objection that punishment in the context of Nonpunitive cooperators will commonly be collective actions suffers from a second-order free trapped in unproductive triads without a punitive rider problem is an important argument, worthy of strategist and with at least one free rider, until serious attention. However, the validity of this punitive strategists become very common. argument depends on the nature of the ancestral In short, the existence of numerous small-scale distribution of opportunities for cooperation, on exchange interactions which fail without punitive other pre-existing components in our evolved cooperators but succeed with them can drive a social psychology, and on the exact nature of the strategy of punitive cooperation to high frequen- computational procedures that implement the cies just by individual selection (in the ordinary punitive psychology. We think, for example, that sense8). Punitive cooperators outcompete both it is highly likely that the ancestral frequency of nonpunitive cooperators and free riders because opportunities for exchange was a declining func- they far more commonly find themselves in tion of the number of individuals involved. That is, potential cooperative interactions without free there were ubiquitous opportunities for dyads, riders and their inhibiting effects. Consequently, frequent opportunities for triads, fewer for tetrads, punitive cooperators are successful at realizing a even fewer for pentads, and so on. In normal far broader range of gains from n-party exchanges hunter–gatherer social , twenty people than are rival strategies. They are not vulnerable to rarely shared joint attention on a common project, the abiding weakness of a strictly conditional while ten people sometimes did, while five people exchange strategy, because interactive contexts often did, and two people commonly did. In such a that initially include one or more free riders size-skewed social ecology, selection for a punitive cannot inhibit them from harvesting valuable joint strategy could easily proceed (1) when the benefit gains. The selection pressure created by free riders to punitive strategists of increased gains from persists across evolutionary time because our participating in n-person exchanges sufficiently evolved psychology of effort minimization makes exceeds the costs of punishment, and (2) when a reversion toward free riding a recurrent phenom- the costs of punishment for punitive strategists are enon in the absence of corrective social feedback sufficiently less in their successful exchange inter- provided by punitive strategists. We think, there- actions than the costs nonpunitive cooperators fore, that this history of selection has constructed a incur from more often finding themselves in failed sophisticated cooperative psychology that includes exchange interactions (Tooby and Cosmides, punitive motivation against exploiters. 2002). Nonpunitive cooperators will frequently Of course, our evolved punitive psychology find themselves in exchange interactions that failed cannot be indifferent to the costs of inflicting because there were no punitive strategists in the punishment, or to the efficiency with which effort interaction to defend against free riders and the devoted to punishment affects the landscape of undercooperation they trigger. Punitive coopera- exploitation. Our evolved punitive psychology tors, of course, always find themselves in interac- should be designed to be sensitive to both of tions that include a punitive cooperator, while these variables. Moreover, the efficiency of punish- nonpunitive cooperators only sometimes do. If all ment will be affected by how many others in exchange interactions involved very large numbers the exchange will participate in the punitive

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 124 J. TOOBY ET AL. enforcement of the exchange. The greater the which coordination can be achieved in collabora- number of individuals that collaborate in punish- tive punishment. ment, the smaller the cost of punishment for each Although our psychology of punitive coopera- individual, and the more efficient punishment will tion evolved in the context of small-scale interac- be. Therefore, our punitive circuits should also be tions, and derives much of its fitness advantage designed to invite others to participate in punitive from the enduring pervasiveness of small-scale enforcement, by preceding punishment with ex- social interaction, the cultural emergence of rarer, posures of exploitation, expressed disapproval or larger scale cooperative contexts need not have outrage, solicitations of social support to respond selected against punitive cooperation designs. One to exploitation, and monitoring for a sufficient reason arises from the fact that individuals are mutuality of sentiment to make punitive action designed to be selective about who they encourage cost-effective.9 Social life is riddled with confiden- to fill the limited social niches in their lives, such as tial complaints about the misbehavior of third friends, cooperators, allies, and exchange partners parties, as individuals sound each other out about (see, e.g. Tooby and Cosmides, 1996). Also, it is shared views, and implicitly about various poten- likely that the tendency to free ride or cheat in tial collective punitive enterprises. Moreover, the some contexts is predictive of the tendency to be size of the exchange interaction will be an exploitive in other contexts (for many reasons, important variable impacting the cost-effectiveness such as the existence of stable individual differ- of individual punitive action. A single blindly ences in impulse control). This means that activated punitive strategist in a large group observations of how others behave in large-scale containing many free riders is unlikely to be exchanges will provide some information about effective, and therefore would be selected against. how they will act in small-scale exchanges. Public Therefore, early in our social evolution, when dereliction in large-scale exchanges should nega- punitive strategists were rare, such a strategy tively influence the willingness of others to include would have only been favored to the extent that highly delinquent individuals in advantageous, circuitry was designed to activate punishment only smaller scale exchanges (see Panchanathan and where individual action would be efficacious}that Boyd, 2004, for a related argument). is, in groups that were sufficiently small. However, as punitive strategists became more common, the Asymmetric beneficiaries, the evolution of leader- costs of punishing would have been shared among ship, and the emergence of morality. There is an them, allowing the threshold of activation to be even more important dimension of n-person relaxed so that punishment was deployed in larger exchanges that makes the evolution of punitive and larger groups. Still, other things being equal, strategies by individual selection nonproblematic. punitive sentiment should be stronger (i.e. the The economic theory of public goods and collec- amount of cost an individual is willing to incur in tive action is based on assuming an idealization in order to punish should be larger) in smaller which the benefits that individuals receive from exchange interactions than in larger exchange collective action are exactly equal. For a broad interactions. Equally, the greater the number of array of ancestral activities, this assumption was individuals who collaborate in punitive enforce- very unlikely to have held. Individual differences in ment, the more readily individuals will become kin arrays, status, alliances, reproductive value, punitively activated, other things being equal. strength, health, and other circumstances would In short, our evolved punitively cooperative commonly have created a spectrum of benefit psychology should be designed to be sensitive to a magnitudes that differed from individual to number of variables related to its function, individual for any given joint enterprise. In such including: the costs of punishment to the punitive a world, there would be no barrier to the evolution strategist; the number of individuals who will of a conditional punitive cooperator strategy that share the costs of punishment; the number of free operated according to the following rule: Initiate riders requiring punishment; how cost-effective and enforce an n-person exchange, using punitive punishment will be in eliminating situations of threats against potential exploiters, whenever the exploitation (including how effective punishment cost of inflicting punishment will be less than the will be in reducing the excess benefits accruing to excess benefit the punitive strategist derives free riders); and the communicative efficiency with than what nonpunitive cooperators derive

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 125

(Tooby and Cosmides, 2002, forthcoming; Price of a coordination), she is better positioned to et al., in preperation). The punitive strategist is become a leader. To the extent that an individual compensated for her excess expenditure of punitive derives an asymmetrically greater benefit from a effort by the fact that she derives a greater benefit given collective effort, she will be more motivated from the interaction than the nonpunitive coop- to expend coordinative effort leading the group to erators do. Rationally or evolutionarily, they achieve the group gain.10 Equally, a leader needs would not be designed to resist the punitive the knowledge and intelligence to make those strategist to the extent that the punitive strategist coordinative decisions that lead to success in a selects n-person exchanges that are in their joint enterprise. That is, she must be cognitively interest, too. In short, by limiting the expenditure equipped to determine what causal steps are of punitive efforts only to those common enter- needed to successfully orchestrate the intended prises that the punitive strategist has a stronger joint outcome. Gains from n-person enterprises relative interest in, the punitive strategy cannot be can be achieved or lost depending on whether the outcompeted by a strategy of second-order free plan corresponds to reality. Finally, a leader needs riding. Such a conditional punitive cooperator to choose, out of the available array of alternative strategy reliably gains net benefits in social plans of action, those coordinated efforts that (1) interaction that alternative strategies such as benefit her, while also (2) sufficiently benefiting a noncooperation, strict conditional cooperation, large enough subset of the potential participants and dispositional free-riding do not. In sum, we that they (3) support the joint project strongly conclude that a psychology of punishment in enough for it to succeed. Social groups are exchange became part of our species-typical design confronted with an indefinitely large array of over evolutionary time, although it is expected to potential joint enterprises, each with its own be differentially activated in different individuals matrix of payoffs. A leader must be able to because of different individual circumstances and compute how different proposals will differentially different developmental trajectories. impact the perceived welfare of potential partici- What does the hypothesized presence of such a pants. A leader who proposes enterprises that psychology predict? Those contributors to an n- differentially benefit herself at the expense of person exchange who stand to gain the most from others will not be supported by others, while an its success will be differentially involved in individual who sacrifices her own interests too organizing and expending coordinative effort: strongly for the enterprise will not be motivated to punishment, encouragement, persuasion, and so expend effort to act as leader. To be successful, a on. Equally, those for whom the cost of inflicting leader must be able to anticipate, perceive and punishment is relatively less (i.e. more powerful or represent the perspective-specific values of the formidable individuals) will also be more likely to interactants, in order to converge on those find themselves in the envelope of conditions collective enterprises with sufficient political sup- where punishment is cost-effective. These provide port. Persuasion involves the ability to discover some of the kernels out of which a theory of the and represent the values of others, and to awaken nature of leadership can be developed. a sense of how personally valuable the anticipated As discussed, leadership spontaneously emerges outcome of the enterprise will be to the audience. from a differential ability to solve the coordination Of course, to the extent that an individual problems involved in n-person exchanges, includ- naturally and inextricably benefits from the same ing the threat of dissolution through individuals set of outcomes that most other group members do choosing to undercontribute. Consequently, one (although perhaps to a greater degree), she is element of leadership is the ability to effectively better positioned to become a leader. Of course, deploy punishment to prevent potential partici- leadership is not an all or nothing matter: All pants from defecting on a joint cooperative participants actively or passively play a role in venture. To the extent that an individual is more directing coordinative effort, instigating or inhibit- formidable (i.e. for whatever reason, can inflict ing action, and otherwise exercising social influ- necessary punishment at lower costs), she is better ence on the direction that the collective enterprises positioned to become a leader. To the extent an we participate in will take. individual is a more effective communicator (and Finally, it is worth briefly considering how these so can communicate the benefits and requirements proposals concerning the evolved psychology of

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 126 J. TOOBY ET AL. n-person exchange provides a theory of the cooperator strategies prosper, the likely effects of transmission of moral precepts in organizations frequency-dependent dynamics on the evolution of and communities. We think that the willingness to the mechanisms, and the necessary additional follow many classes of moral rules within com- cognitive abilities and motivational circuits hu- munities and organizations is implicitly treated by mans must have for conditional cooperation, the mind as a form of n-person exchange. punishment, leadership, and morality to emerge. Logically speaking, there need be no relationship Nevertheless, we think that this treatment can between whether some individuals follow a rule provide some insights into how humans are and whether others do, if the goal is to be ethical. psychologically designed behave in organizations, Yet, we implicitly treat many rules as if following and what variables will govern the dynamics, them were somehow conditional on others’ con- success, and failure of various organizational duct. Many people feel sentiments according to the forms. following social exchange logic: I will give up the benefits of violating this moral rule if others in my Acknowledgements social world do. If I followed the rule, and you did Thanks to Pat Barclay, Will Brown, Oliver Curry, Ed Hagen, not, I have been cheated by you. The more others Lin Ostrom, Jade Price, Steve Rothstein, Don Symons, cheat on a rule I follow, the more exploited I feel, Masanori Takezawa, Robert Trivers, members of the Beha- vioral Science Group at the Santa Fe Institute, and members of and the more tempted I am to discontinue the Center for Evolutionary Psychology Research Seminar. following the rule when it is costly to do so. Funding was provided by the Indiana University Workshop in Whether or not we act on it, we feel a link between Political Theory and Policy Analysis, and by a grant to the Santa Fe Institute from the James S. McDonnell Foundation- our motivation to follow moral rules and others’ 21st Century Collaborative Award for the Study of Complex adherence to them. The hypothesis that morality is Systems. treated psychologically as a collective exchange also provides an explanation for our otherwise irrational response to hypocrisy. Although the value of an argument is logically unrelated to who NOTES says it, individuals act as if their willingness to adopt a moral proposal depends on whether the 1. It is better not to think of economic rationality as an proposer follows it herself. Nothing destroys a explanation for behavior}that is, as a set of causes of anything. That is, theories of economic rationality leader’s credibility faster than the discovery that are high level post hoc descriptive idealizations of she has been a hypocrite. It is as if a cheater outputs, rather than a model of a real process that proposes a social exchange: Our psychology is produces or explains behavior. As economics, orga- designed to avoid entering into exploitive relation- nizational behavior, and psychology mature and ships, and adopting a moral rule from one who is integrate, the ground level theories will be detailed descriptions of the information-processing architec- not following it maps into our exchange psychol- tures of the mechanisms in the human brain, and how ogy as an instance of allowing oneself to be they interact in sets in structured environments. cheated. This hypocrisy circuit makes no sense Using rationality as an account of choice is parallel logically, but makes evolutionary psychological to using animism as an account for the ability of sense to the extent that morality in organizations animals to move}admirably descriptive, while fail- ing to track or connect with the science of actual and communities is one expression of our ex- causation. change psychology. 2. One can view the phenomenon of increasing con- servatism with increasing scale either negatively or positively, as either cognitive inertia or the coordi- native facilitation of pre-existing coordinative struc- tures. On the plus side, far more large-scale CONCLUSIONS cooperation is realized than otherwise would exist if humans had to build exchange structures de novo for The picture of n-person exchange psychology that each individual opportunity for an interaction to we have sketched here is of course highly realize a gain in trade. On the downside, the fact that simplified both computationally and game theore- the revision of pre-existing practices is increasingly retarded the larger the interaction structure means tically. For example, we have not been able to that larger organizations increasingly grandfather in discuss a number of game theoretic complexities obsolete and nonfunctional features. Conversely, the involving the strategy sets against which punitive smaller the interaction set, the more dynamic it will

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 127

be, and the more rapidly it will be tailored to meet evolved infrastructure out of which the punitive changing conditions. The larger the structure, the sentiment system evolved. Indeed, to the extent that more processes of cultural transmission will be our evolved psychologies are designed to recognize important, while the smaller the structure, the more that free riding inhibits n-person exchanges and game-theory like interactive responses to existing the gains they generate (while simultaneously bene- conditions will predominate. This principle explains fiting the free rider) the anger system would why small, young companies more effectively exploit process this as a situation in which the free rider is new economic opportunities, despite the capitaliza- trading off the welfare of others for her welfare. To tion advantages large companies enjoy. the extent that the exploiter is inflicting unacceptably 3. By parallel logic, it should be easier to grow a high costs in pursuit of insufficient personal coordinated n-person exchange structure from a gains, this should trigger anger at exploitation. smaller to a larger size than to create it de novo at a That is, anger and punitive sentiment are closely large size. related functional systems, using much of the same 4. This is true even of riots, which usually outgrow the infrastructure. initial focus or pretext and become a wide-ranging 8. Individual selection and group selection have come to opportunity to realize those joint values of the rioters acquire such a broad variety of meanings that it is that are usually individually deterred by a lack of difficult to speak briefly about them with any clarity. sufficiently coordinated collective effort. David Sloan Wilson, for example, uses such a broad 5. Socially marginal individuals should be more eager to definition of group selection that dyadic reciproca- covertly feed disputes among others. tion and kin selection both become examples of 6. Cults, for example, place a strong emphasis on ‘group selection’ because they involve interactions in cutting ties to families and other outsiders (e.g. ‘groups’ (Wilson, 1980). We prefer to restrict the term Lalich, 2004), and basic training isolates new recruits group selection to those cases where the higher level from family members for the first several months of (‘group’ derived) component of the fitness of the training. organism derives from an organism’s stable member- 7. We consider anger to be the expression of an evolved ship in a single group for a majority of its life history, emotion program (a functionally structured neuro- rather than from its transient participation in computational system) whose design features and thousands of different ‘groups’ or (as we would put subcomponents evolved to advantageously regulate it) interactions. In our view of the evolution of n- thinking, motivation, and behavior in the context of person exchange, different individuals (embodying resolving conflicts of interest in favor of the angry different designs) accumulate fitness differences be- individual. Two negotiating tools regulated by this cause of how they act across large numbers of system are the threat of inflicting costs (aggression) different social interactions. That is, a person might and the threat of withdrawing benefits (the down- engage in 20 dyadic interactions in a day, and 8 regulation of cooperation). Humans have a system triadic interactions, and 1 twenty-person interaction. that, in each individual, recognizes the welfare trade- Confusion arises by labeling every social interaction a off ratio expressed in the actions another party takes group. with respect to oneself. Anger is conceptualized as a 9. A strategy is defined computationally by the detailed mechanism whose functional product is the recali- specification of a behavior control program, includ- bration in the mind of another of this other person’s ing how it represents conditions in order to regulate welfare trade-off ratio with respect to oneself (or behavior. The nature of the implementation matters: other valued agent). That is, the goal of the system is Consider, for example, a program design for a to change the targeted persons’ disposition to make punitive strategist that sums both punitive effort welfare trade-offs so that they more strongly favor and cooperative effort together into one unitary the angered individual (or their group) in the present account as its definition of the contribution it and the future. As in contests, the target of considers either self or others makes to the collective anger may relinquish a contested resource, or may enterprise. This particular design will not suffer from simply in the future be more careful to help or to a second-order free rider problem. The punitive avoid harming the angered individual. In cooperative strategist will contribute less nonpunitive effort to relationships, where there is the expectation that the make up for the additional punitive effort it is cooperative partner will spontaneously take the contributing. Alternatively, it could insist that non- welfare of the individual into account, the primary punitive cooperators make a larger nonpunitive threat from the angered person that potentially contribution to make up for the additional punitive induces recalibration in the targeted individual is effort it is expending. Indeed, the second-order free the signaled possibility of the withdrawal of future rider problem only exists to the extent that partici- help and cooperation if the welfare trade-off ratio is pants do not compute punishment as itself a not modified. In the absence of cooperation, the contribution. But punishment is in reality a contribu- primary threat is the infliction of damage. Concepts tion, and punitive strategists (as well as rational that are anchored in the internal psychological cooperators, if any) will view it as such. In any case, a variable welfare trade-off ratio include respect, con- mutant design that considers punishment a coopera- sideration, deference, status, rank, and so on. An- tive contribution cannot be exploited by second-order ger is an ancient system, and provides much of the free riders. It may prosper, or not, depending on the

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) 128 J. TOOBY ET AL.

details of how conditional cooperation is implemen- Fehr E, Ga¨chter S. 2000. Cooperation and punishment ted in nonpunitive cooperators. in public goods experiments. American Economic 10. It is a commonplace of democratic politics that Review 90: 980–994. many proposals that would modestly benefit large Fischbacher U, Ga¨chter S, Fehr E. 2001. Are people majorities of the population are not acted on, in conditionally cooperative? Evidence from a public favor of other proposals that strongly benefit far goods experiment. Economics Letters 71: 397–404. smaller fractions of the population. Fiske, AP. 1991. Structures of Social Life: The Four Elementary Forms of Human Relations. Free Press: New York. Henrich J. 2004. Cultural group selection, co- evolutionary processes and large-scale cooperation. Journal of Economic Behavior and Organization 53: REFERENCES 3–35. Henrich J, Smith N. 2004. Comparative experimental Axelrod R, Hamilton WD. 1981. The evolution of evidence from Machiguenga, Mapuche, Huinca and cooperation. Science 211: 1390–1396. American populations. In Foundations of Human Baron-Cohen S, Leslie AM, Frith U. 1985. Does the Sociality, Henrich J, Boyd R, Bowles S, Camerer C, autistic have a ‘theory of mind’? Cognition 21: 37–46. Fehr E, Gintis H (eds). Oxford University Press: Boyd R, Gintis H, Bowles S, Richerson PJ. 2003. The Oxford; 125–167. evolution of altruistic punishment. Proceedings of the Hoffman E, McCabe KA, Smith VL. 1998. Behavioral National Academy of Sciences (USA) 100: 3531–3535. foundations of reciprocity: experimental economics Boyd, R, Richerson PJ. 1992. Punishment allows the and evolutionary psychology. Economic Inquiry 36: evolution of cooperation (or anything else) in sizable 335–352. groups. Ethology and Sociobiology 13: 171–195. Kurzban R, McCabe K, Smith VL, Wilson BJ. 2001a. Brase GL. 1998. Human reasoning about social groups: Incremental commitment and reciprocity in a real time evidence of evolved mechanisms for a coalitional public goods game. Personality and Social Psychology psychology. Dissertation Abstracts International: Bulletin 27: 1662–1673. Section B: The Sciences and Engineering 58(9-B): Kurzban R, Tooby J, Cosmides L. 2001b. Can race be 5150 (UMI# 9809613). erased?: Coalitional computation and social categor- Brewer MB. 1991. The social self: on being the same and ization. Proceedings of the National Academy of different at the same time. Personality and Social Sciences 98(26), 15387–15392 (December 18, 2001; Psychology Bulletin 17: 475–482. Epub 2001 Dec 11. PMID: 11742078). Cosmides L. 1989. The logic of social exchange: has Lalich J. 2004. Bounded Choice: True Believers and natural selection shaped how humans reason? Studies Charismatic Cults. University of California Press: with the Wason selection task. Cognition 31: 187–276. Berkeley. Cosmides L, Tooby J. 1989. Evolutionary psychology Ledyard JO. 1995. Public goods: a survey of experi- and the generation of culture, Part II. Case study: a mental research. In The Handbook of Experimental computational theory of social exchange. Ethology Economics, Kagel JH, Roth AE (eds). Princeton and Sociobiology 10: 51–97. University Press: Princeton; 111–194. Cosmides L, Tooby J. 1992. Cognitive adaptations for Lubell M, Scholz JT. 2001. Cooperation, reciprocity and social exchange. In The Adapted Mind: Evolutionary the collective-action heuristic. American Journal of Psychology and the Generation of Culture, Barkow JH, Political Science 45: 160–178. Cosmides L, Tooby J (eds). Oxford University Press: Madison J. 1788/1982. Federalist paper no. 10. In The New York; 163–228. Federalist Papers (2nd ver), Hamilton A, Jay J, Cosmides L, Tooby J. 2000. Consider the source: the Madison J (eds). Bantam Classics; Reissue edition: evolution of adaptations for decoupling and metar- New York. epresentation. In Metarepresentations: A Multidisci- Masclet D, Noussair C, Tucker S, Villeval M. 2003. plinary Perspective, Sperber D (ed.). Oxford Monetary and nonmonetary punishment in the University Press: New York; 53–115. voluntary contributions mechanism. American Cosmides L, Tooby J. 2005. Neurocognitive adaptations Economic Review 93: 366–380. designed for social exchange. In The Handbook of Maynard Smith J. 1982. Evolution and the Theory of Evolutionary Psychology, Buss DM (ed.). Wiley: New Games. Cambridge University Press: Cambridge, UK. York. Olson M. 1965. The Logic of Collective Action: Public Darley JM, Latane B. 1968. Bystander intervention in Goods and the Theory of Groups. Harvard University emergencies: diffusion of responsibility. Journal of Press: Cambridge. Personality and Social Psychology 8: 377–383. Orbell J, Dawes R. 1991. A ‘cognitive miser’ theory of Dumas A. 1844/2001. The Three Musketeers. Random cooperators’ advantage. American Political Science House: New York. Review 85: 515–528. Ehrhart K-M, Keser C. 1999. Mobility and Cooperation: Orbell J, Dawes R. 1993. Social welfare, co- On the Run. Scientific Series. CIRANO: Montreal. operators’ advantage, and the option of not Fehr E, Fischbacher U. 2003. The nature of human playing the game. American Sociological Review 58: altruism. Nature 425: 785–791. 787–800.

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006) COGNITIVE ADAPTATIONS FOR N-PERSON EXCHANGE 129

Ostrom E. 1990. Governing the Commons: The Evolution in adaptive problems. Center for Evolutionary Psy- of Institutions for Collective Action. Cambridge chology Technical Report 92-1 (Reprinted in: Tooby J, University Press: New York. Cosmides L. (in press) Evolutionary Psychology: Ostrom E. 1998. A behavioral approach to the rational Foundational Papers. With a foreword by Steven choice theory of collective action. Presidential Pinker. MIT Press: Cambridge). Address, American Political Science Association, Tooby J, Cosmides L. 1996. Friendship and the Bank- 1997. American Political Science Review 92: 1–22. er’s Paradox: Other pathways to the evolution of Panchanathan K, Boyd R. 2004. Indirect reciprocity can adaptations for altruism. In Evolution of Social stabilize cooperation without the second-order free Behaviour Patterns in and Man, Runciman rider problem. Nature 432: 499–502. WG, Maynard Smith J, Dunbar RIM (eds). Oxford Price ME. 2003. Pro-community altruism and University Press: Oxford; 119–143. social status in a Shuar village. Human Nature 14: Tooby J, Cosmides L. 2000. Cognitive adaptations for 191–208. kin-based coalitions: human kinship systems at the Price ME. in press A. Monitoring, reputation and intersection between collective action and kin selec- ‘greenbeard’ cooperation in a Shuar work team. tion. Current Anthropology 41: 803–804. Journal of Organizational Behavior. Tooby J, Cosmides L. 2002. The evolution of collective Price ME. in press B. Punitive sentiment among the action: an adaptationist dissection. Human Behavior Shuar and in industrialized societies: cross-cultural and Evolution Society Meetings, Rutgers University, similarities. Evolution and Human Behavior. Newark, NJ, 19–23 June. Price ME, Cosmides L, Tooby J. 2002. Punitive Tooby J, Cosmides L. 2005. The functional architecture sentiment as an anti-free rider psychological device. of human motivation. Paper presented at the Human Evolution and Human Behavior 23: 203–231. Behavior and Evolution Society Meetings, Austin, TX. Price ME, Tooby J, Cosmides L. in preparation. The Tooby J, Cosmides L. forthcoming. Human coalitional evolutionary psychology of work teams. psychology. Oxford Handbook of Evolutionary Psy- Sidanius J, Pratto F. 1999. Social Dominance: An chology. Dunbar RIM, Barrett L (eds). Oxford Intergroup Theory of Social Hierarchy and Oppression. University Press: New York. Cambridge University Press: New York. Tooby J, Cosmides L, Barrett HC. 2005. Resolving the Stone V, Cosmides L, Tooby J, Kroll N, Knight R. debate on innate ideas: learnability constraints and the 2002. Selective impairment of reasoning about social evolved interpenetration of motivational and concep- exchange in a patient with bilateral limbic system tual functions. In The Innate Mind: Structure and damage. Proceedings of the National Academy of Contents, Carruthers P, Laurence S, Stich S (eds). Sciences 99(17):11531–11536. Oxford University Press: Oxford. Tiger L. 2004. Men in Groups (3rd edn). Transaction Trivers RL. 1971. The evolution of reciprocal altruism. Publishers: New York. The Quarterly Review of Biology 46: 35–57. Tooby J, Cosmides L. 1992a. The psychological Wilson DS. 1980. The Selection of Populations and foundations of culture. In The Adapted Mind: Evolu- Communities. Addison-Wesley: Menlo Park. tionary Psychology and the Generation of Culture, Williams GC. 1966. and Natural Selection. Barkow JH, Cosmides L, Tooby J (eds). Oxford Princeton University Press: Princeton. University Press: Oxford; 19–136. Yamagishi T. 1986. The provision of a sanctioning Tooby J, Cosmides L. 1992b. Ecological rationality and system as a public good. Journal of Personality and the multimodular mind: grounding normative theories Social Psychology 51: 110–116.

Copyright # 2006 John Wiley & Sons, Ltd. Manage. Decis. Econ. 27: 103–129 (2006)