Intuition, deliberation, and the evolution of cooperation

1 intuition , deliberation , and the evolutionof cooperationAdam Beara,1and David G. Randa,b,c,1aDepartment of Psychology, Yale University, New Haven, CT 06511;bDepartment of Economics, Yale University, New Haven, CT 06511; andcSchool ofManagement, Yale University, New Haven, CT 06511 Edited by Susan T. Fiske, Princeton University, Princeton, NJ, and approved November 24, 2015 (received for review September 5, 2015)Humans often cooperate with strangers, despite the costs involved. Along tradition of theoretical modeling has sought ultimate evolution -ary explanations for thisseemingly altruistic behavior. More recently,an entirely separate body of experimental work has begun to inves-tigate cooperation s proximate cognitive underpinnings using a dual-process framework: Is deliberative self-control necessary to reign inselfish impulses, or does self-interested deliberation restrain an intu-itive desire to cooperate?

2 Integrating these ultimate and proximateapproaches, we introduce dual-process cognition into a formal game-theoretic model of the evolution of cooperation . Agents play pris-oner s dilemma games, some of which are one-shot and others ofwhich involve reciprocity. They can either respond by using a gener-alized intuition , which is not sensitive to whether the game is one-shot or reciprocal, or pay a (stochastically varying) cost to deliberateand tailor their strategy to the type of game they are facing. We findthat, depending onthe level of reciprocity and assortment, selectionfavors one of two strategies: intuitive defectors who never deliber-ate, or dual-process agents who intuitively cooperate but sometimesuse deliberation to defect in one-shot games.

3 Critically, selectionnever favors agents who use deliberation to override selfish impulses: deliberation only serves to undermine cooperation with , by introducing a formal theoretical framework for exploringcooperation through a dual-process lens, we provide a clear answerregarding the role of deliberation in cooperation based on evolu-tionary modeling, help to organize a growing body of sometimes-conflicting empirical results, and shed light on the nature of humancognition and social decision process| cooperation |evolutionary game theory|prisoner s dilemma|heuristicsCooperation, where people pay costs to benefit others, is adefining feature of human social interaction. However, ourwillingness to cooperate is puzzling because of the individualcosts that cooperation entails.

4 Explaining how the selfish process of evolution could have given rise to seemingly altruisticcooperation has been a major focus of research across the nat-ural and social sciences for decades. Using the tools of evolu-tionary game theory, great progress has been made in identifyingmechanisms by which selection can favor cooperative strategies,providing ultimate explanations for the widespread cooperationobserved in human societies (1).In recent years, the proximate cognitive mechanisms underpin-ning human cooperation have also begun to receive widespreadattention. For example, a wide range of experimental evidencesuggests that emotion and intuition play a key role in motivatingcooperation (2 5). The dual-process perspective on decision making(6 8) offers a powerful framework for integrating these observa-tions.

5 In the dual-process framework, decisions are conceptualizedas arising from competition between two types of cognitive pro-cesses: (i) automatic, intuitive processes that are relatively effortlessbut inflexible; and (ii) controlled, deliberative processes that arerelatively effortful but flexible. In many situations, intuitive anddeliberative processes can favor different decisions, leading to innerconflict: Rather than being of a single mind, people are torn be-tween competing the widespread attention that dual-process theories havereceived in the psychological and economic sciences (includingincorporation into formal decision making models; refs. 9 11); theexistence of related discussion in the theoretical biology literatureregarding error management (12 14), tradeoffs between fixed andflexible behaviors (15 18), and cultural evolution and norm in-ternalization (2, 19, 20).

6 And a long interdisciplinary tradition ofarguments suggesting that strategies developed in repeated inter-actions spill over to influence behavior in one-shot anonymoussettings (21 25), the dual-process framework has been almost en-tirely absent from formal models of the evolution of evolutionary game theory models of cooperation focuson behavior, rather than the cognition that underlies , these models do not shed light on when selection mayfavor the use of intuition versus deliberation , or which specificintuitive and deliberative responses will be favored by this paper, we build a bridge between ultimate and proximatelevels of analysis to address these questions, introducing an evo-lutionary game-theoretic model of cooperation that allows fordual-process agents.

7 These agents interact in a varied social en-vironment, where interactions differ in the extent to which currentactions carry future consequences. To capture the tradeoff be-tween flexibility and effort that is central to many dual-processtheories, we allow our agents to either (i) use an intuitive responsethat is not sensitive to the type of interaction currently faced; or(ii) pay a cost to deliberate, tailoring their action to the details ofthe current role of intuition versus deliberation in human cooperationhas received widespread attention from experimentalists acrossthe behavioral sciences in recent years. Yet a formal theoreticalframework for addressing this question has been absent. Here,we introduce an evolutionary game-theoretic model of dual-process agents playing prisoner s dilemma games.

8 We find that,across many types of environments, evolution only ever favorsagents who (i) always intuitively defect, or (ii) are intuitively pre-disposed to cooperate but who, when deliberating, switch to for why we should expect deliberation to promoteselfishness rather than cooperation and unifies apparently con-tradictory empirical results regarding intuition and contributions: and designed research, performed research, analyzeddata, and wrote the authors declare no conflict of article is a PNAS Direct available online through the PNAS open access deposition: Code implementing the model in MATLAB is available whom correspondence may be addressed. Email: or article contains supporting information online 941|PNAS|January 26, 2016|vol.

9 113|no. by guest on November 28, 2021 We then use this framework to explore the consequences ofreciprocity and assortment (26, 27), two of the most widely studiedmechanisms for the evolution of cooperation . We ask when (andto what extent) agents evolve to pay the cost of deliberation ; whenevolution favors intuitive responses that are selfish versus cooper-ative; and whether deliberation serves to increase or decrease socialwelfare. In doing so, we provide a formal theoretical framework toguide the emerging body of empirical work exploring proso-ciality from a dual-process perspective, and provide insight into thecognitive underpinnings of human are two key dimensions on which our model differs fromtypical models of the evolution of cooperation : (i) in each genera-tion, agents play more than one type of game.

10 And (ii)agentsneednot have a single fixed strategy, but can engage in costly deliberationto tailor their response to the type of game they are respect to multiple game types, our agents face both one-shot anonymous prisoner s dilemma (PD) games (which occurwith probability 1 p) and PDs where reciprocal consequencesexist (which occur with probabilityp). In the one-shot PDs,agents can cooperate by paying a costcto give a benefitbto theirpartner, or defect by doing nothing. In the games with reciprocalconsequences, we capture the core of reciprocity [be it via re-peated interactions, reputation effects, or sanctions (1)] bymodifying the off-diagonal elements of the PD payoff structure:When exploitation occurs, such that one player defects while theother cooperates, the benefit to the defector is reduced (due to, , lost future cooperation , damaged reputation, or materialpunishment), as is the cost to the cooperator (due to, ,switching to defection, improved reputation, or material rewards).