Do we need complex cognition for the evolution of cooperation? Implications of Conditional Movement C. Athena Aktipis John Pepper corresponding author email: [email protected] University of Arizona Department of Ecology and Evolutionary Biology BioSciences West 310 Tucson, AZ 85721

Abstract Standard evolutionary and economic approaches to understanding cooperation assume that individual self-interest is the natural state of the world. In that view, cooperation is only viable among individuals with sufficient cognitive complexity to overcome the ‘natural’ state of defection through such abilities as individual recognition, memory, punishment, commitment and other incentive and enforcement systems. The present paper challenges the view that complex cognitive abilities are necessary for the evolution of cooperation by describing the findings of two models of the evolution of cooperation: the Walk Away and environmental feedback models. In these models, individuals simply leave regions in which they receive low returns. Because the presence of defectors reduces the quality of the local environment, regions with defectors are less stable than regions of cooperators. The individual level behavior in these models generates aggregate dynamics that promote positive assortment and selection for cooperation via group selection. The basic findings of the Walk Away and environmental feedback models suggest that complex cognitive abilities may not be necessary for the evolution of cooperation and that cooperation may be the ‘state of nature’ in a wider variety of contexts than previously assumed.

Introduction The notion that conflict is the natural state of the world has a long and interdisciplinary history, engaging Philosophers, Behavioral Scientists and Evolutionary Biologists. The question of whether the state of nature is inherently one of conflict is both a longstanding Philosophical question, as well as a testable hypothesis. Is the natural state of the world inherently Hobbesian, “a condition of war of every one against every one,” (1651) or naturally benevolent, as Rousseau (1755) argues?

Evolution is an inherently competitive process, and it is often assumed that this Darwinian process should result in individuals that are inherently competitive rather than cooperative. On this view, cooperation1 is the exception rather than the rule – cooperation is something that must be explained with concepts such as kin selection and reciprocity (Hamilton 1964a; Hamilton 1964b; Dawkins 1976). Likewise in Economic, self-interest is the basic assumption of the ubiquitous rational actor model. In these Economic models, cooperation is explained by including ‘other oriented preferences’ into an otherwise selfish utility function. Is this view of nature “red in tooth and claw,” reasonable given what we know of the natural world? Here, we argue that one very simple feature of the natural world – the ability of many organisms to leave low-quality areas – has important implications for the debate about whether cooperation may be a natural state of the world. In particular, when individuals can both affect the quality of their environments and leave degraded environments, this produces dynamic social structures that promote the evolution of cooperation (C. Athena Aktipis 2008; C. Athena Aktipis 2004; J. W Pepper and B. B Smuts 2002; John Pepper; John W Pepper and Barbara Smuts 1999). Because this ability to leave low-quality environments is both evolutionarily ancient and widespread, it forces us to consider the possibility that cooperation may be the natural state for a wide variety of organisms. This includes very simple organisms such as bacteria, which can leave degraded environments but are incapable of cognition or complex behaviors. On this view, cooperation can emerge and be sustained without complex cognitive abilities. Our view – that cooperation is viable without cognition – runs counter to the view expressed, both implicitly and explicitly, in a large body of work exploring cooperation in individuals with cognitive abilities such as individual recognition, memory for past interactions, punishment, others. For example, in Robert Axelrod’s famous book, The Evolution of Cooperation (1984), he states “the ability to recognize the other player from past interactions, and to remember the relevant features of those interactions, is necessary to sustain cooperation” (p. 139). In a recent review paper of cooperation in humans, Eric Smith (2003) notes that humans solve the problem of cooperation by using a variety of complex strategies including “communicating about options and preferences, socially transmitting norms and other codified information, monitoring the behavior of others, imposing punishment for selfish behavior, and dispensing selective incentives for cooperative or prosocial behavior” (pp.401-402). Certainly these abilities exist in humans and appear to promote cooperation, but are they necessary for cooperation? The relative simplicity of the conditional movement and the apparent complexity of human social life might make such a rule seem an unlikely strategy for human cooperation. Clearly, humans do more than simply leave low-quality social environments and insufficiently cooperative partners. The complexity of human social life is readily apparent to all of us, perhaps contributing to the assumption that human cooperation is complex and informationally intensive.

1

For the present purpose we define cooperation as the transmission of benefits to others.

Conditional movement behavior is certainly unlikely to capture the subtleties of human interaction. Nonetheless, empirical evidence exists that humans are cooperative (Pepper 2007; Aktipis and Kurzban 2004) and do use something akin to the conditional movement in economic games (Barclay and Willer 2007; Orbell, Schwartz-Shea, and Simmons 1984; Boone and Macy 1999) and evolutionarily ancient origins of conditional movement suggest that this strategy may have important implications for our understanding the foundations of cooperation and social behavior in a variety of species, humans included.

Conditional Movement and Cooperation In this paper, we describe a route to the evolution of cooperation based on a trait that is evolutionarily ancient and extremely simple. In particular, we argue that the simple capacity to leave low-quality environments generates dynamics that support cooperation. We discuss the findings of two closely related models, the Walk Away model of Aktipis (2004) and the environmental feedback model of Pepper (Pepper and Smuts 2000, 2002; Pepper 2007). These two simple models of contingent movement demonstrate that cooperation can be stable and robust without complex cognition. In both Walk Away and environmental feedback models, individual leave regions containing defectors. This promotes the evolution of cooperation by generating systematic movement away from regions containing defectors, causing positive assortment and the instability in regions with defectors (e.g., uncooperative groups). The Walk Away and environmental feedback models described here challenge the notion that individualism is the ‘natural’ state of the world and that complex cognitive abilities are needed for cooperation to be viable. These models demonstrate that cooperation is robust and stable when agents use a simple strategy based on basic foraging principles. Because cooperation is promoted when individuals follow the evolutionarily ancient rule of leaving low-quality environments, cooperation may be the state of nature in a wider variety of context that previously considered. In these models, cooperation does not require complex cognitive abilities, only the capacity to respond to the local environment. In this paper we note that: 1) The capacity to leave low-quality environments is evolutionarily ancient. 2) When individuals can leave low-quality social environments while influencing environmental quality, this generates environmental feedback (i.e., defectors degrade their local environments resulting in systematic emigration from regions with defectors). 3) The Walk Away rule is a social generalization of a basic foraging rule to leave low-quality environments. 4) Systematic movement generates positive assortment and differential group dissolution, which promote group selection and the stability of cooperation.

From these basic assumptions/findings, we argue that cooperation can be considered a ‘state of nature’ for organisms capable of leaving low-quality areas, and that the evolution of cooperation does not require complex cognition.

Simple foraging rules generate environmental feedback Conditional movement is well documented in a wide variety of species, especially in the context of foraging behavior. The ability to leave a foraging patch or location when returns are low is a basic strategy for optimizing foraging behavior and evolutionary fitness. This requires both conditional movement and the ability to sense local patch quality or its proxies (Stephens et al. 2007). Although complex foraging trade-offs occur for some organisms, conditional movement based on local environmental quality appears to be almost as old as life itself, as it is common in prokaryotes (Glagolev 1984) and may pre-date eukaryotic life. In animals, foraging interacts strongly with social organization and behavior. Both group size and group membership are influenced by, and also influence, foraging and movement behaviors (Waite & Field 2007). When organismal behavior both effects, and is effected by, local environmental conditions, the elements are in place for “environmental feedback”. The earliest models of this dynamic were detailed enough to explicitly include the ecological mechanisms by which feeding behaviors effected local patch quality, which in turn influenced forager movements. (Pepper & Smuts 2000). (Here rapacious foragers over-exploited food resources, leading to forager dispersal.) Because these mechanistic details can vary for different species and environments, while the same dynamics come into play, later models of environmental feedback omitted detailed mechanisms, and simply assumed that individuals vary in their effect on local environmental quality, while sharing in common the tendency to leave low-quality patches faster than high-quality patches (Pepper & Smuts 2002, Pepper 2007). These environmental feedback models have demonstrated that cooperators affect their local environments in ways that generate positive assortment. This same process is at work in the Walk Away model: groups with cooperators are more stable than groups with defectors because defectors decrease the payoffs to partners or group members, making it more likely that others will ‘walk away’ because of low returns.

Walk Away generates social dynamics promoting cooperation Although developed independently, the Walk Away and environmental feedback models share the basic feature that individuals leave low-quality environments. Walk Away can, in this sense, be considered a social generalization of the environmental feedback model. Although the Walk Away models and environmental feedback models implement the

same conditional movement rule, Walk Away was designed specifically as a model of social interactions in partnerships (Aktipis 2004) and groups (Aktipis 2008). In Walk Away models, individuals directly perceive and respond to the level of cooperation of those they interact with. Under the dyadic Walk Away rule individuals leave a patch immediately upon interacting with a defector, without any role for memory or recognition. Similarly, in the group-wise Walk Away model (C. Athena Aktipis 2008), individuals leave groups that are insufficiently cooperative. As in environmental feedback models, Walk Away models are spatial, so movement away from the current location provides a way for individuals to escape from uncooperative groups or partners. In the Walk Away models, individuals interact in a prisoner’s dilemma game (Aktipis 2004) or public goods game (C A Aktipis). Individuals reproduce and die based on their payoffs from these interactions. Within a particular group/partnership, defectors always have a higher payoffs, giving defectors (or, low investors in the continuous investment model) an evolutionary advantage within each group/partnership. In dyadic simulations (Aktipis 2004), partnerships with defectors are unstable because Walk Away cooperators leave after one interaction with a defector. This leads defectors to have fewer interactions (than cooperators) because they degrade their social environments, ‘destroying’ the social relationship with a potential partner by defecting. In these dyadic simulations, Walk Away outperformed Tit-for-Tat and PAVLOV under a variety of parameter values and was able to invade populations of defectors with very few initial Walk Away individuals (much fewer that PAVLOV and Tit-for-Tat). In the group-wise model (Aktipis 2008), similar dynamics emerge, but the dissolution of groups is not as rapid as the dissolution of partnerships in the dyadic model. Within each group, defectors have higher payoffs than cooperators, leading them to reproduce more quickly than cooperators. However, as these defectors increase in frequency within the group, they decrease the payoffs for all group members. When individuals begin to ‘walk away’ because of low payoffs, they can form many small groups on the periphery. These new groups typically vary in their cooperativeness and the more cooperative groups grow more quickly because higher payoffs are generated (on average) in these groups. This increases the level of cooperation in the overall population as the variation between groups in overall cooperativeness generates differential fitness between groups. This leads to a situation in which groups with more cooperators are more stable, while groups with defectors are less so. This differential stability of cooperative groups promotes the evolution of cooperation because it allows cooperators to interact with one another over many time periods (promoting positive assortment, as in the environmental feedback model). Groups with more cooperators last longer and have higher payoffs than groups with defectors, giving cooperators greater opportunities to reproduce than defectors (despite the within-group advantage of defectors). In the Walk Away model, groups stay stable as long as individuals are cooperative enough. As within-group selection favors defectors within groups, these groups become less stable because individuals in these uncooperative groups begin to ‘walk away.’

When these groups dissolve, new groups get formed on the periphery, some of which are more cooperative that the initial group. More cooperative groups then increase in size more quickly than the less cooperative groups, ratcheting cooperation to higher and higher levels despite the within-group advantage of defectors (Figure 1).

Figure 1. This schematic illustrates Simpson’s Paradox: the level of cooperation within each group decreases over time while the average level of cooperation increases. This is driven by the constant creation of new groups through agent movement and regrouping as well as the faster rate of growth for the groups that are most cooperative. Group size is indicated by the thickness of the black lines and the thin gray lines show cases where new groups are created because of contingent movement (as a result of agents’ thresholds being reached).

Partner Choice and Assortment The importance of choice or voluntary participation in partnerships and groups is an ancient issue in Philosophy and a central issue in social contract theory. On Socrates’ account of social contract theory, choosing not to leave a particular social group entails an implicit contract to follow the rules and laws of that society (Plato). More recently, Philosophers such as Skyrms (2003) have noted that the choice of whether or not to engage in social interactions can promote interactions between ‘like-minded’ individuals. This interaction of cooperators with one another is the essence of positive assortment. It has recently been suggested that assortment underlies all models of the evolution of cooperation, including kin selection, group selection, reciprocity models and more complex abilities (Aktipis 2008; Fletcher and Doebeli 2009). When individuals engage in behaviors that promote assortment, this can increase selection for cooperation.

Partner choice models have examined decision rules of varying complexities, from very simple decision rules that respond to the last behavior of the partner (Aktipis 2004) or the state of the local environment (Pepper & Smuts 1999; Pepper & Smuts 2002), to more complex rules with complex memory (Yamagishi, Hayashi, and Jin 1994; Yamagishi and Hayashi 1996; Vanberg and Congleton 1992) (Vanberg & Congleton 1992; Yamagishi et al 1994) or both contingent cooperation and contingent movement (Hamilton & Taborsky 2005). Experimental work also suggests that partner choice is likely to be an important component of human cooperative decision making (Barclay & Willer 2007; Boone & Macy 1999; Orbell et al 1984). The Walk Away model demonstrates that cognition is not necessary for partner choice to be effective. In the Walk Away model, partner ‘choice’ is simply the act of leaving the current partner or group. Individuals in the Walk Away model do not have the ability to choose which partner or group to join once they leave, they simply move randomly until they encounter a new partner or group. This ability to leave uncooperative groups does not require complex abilities. All that is required is the ability to detect and leave lowquality social environments.

Does complexity serve cooperation or competition? We began this paper by asking whether complex cognitive abilities are necessary for the evolution of cooperation, noting that this question is intimately connected to the question of whether cooperation or individualism/defection is the natural state of the world. If defection is the natural state of the world, complex abilities are not needed to overcome the natural pull towards individual-level goals. In contrast, if cooperation is the natural state of the world, complex cognitive capacities are not necessary for cooperation. In this paper we describe the findings of two simple contingent movement models which demonstrate the viability of cooperation without cognitive complexity. However, in a world in which cooperation is that state of nature (Rousseau, 1755), complex cognitive abilities may enable organisms to more effectively exploit one another. Indeed, it has been suggested by many others that human cognitive complexity may have evolved to enable more effective social manipulation. In this view, human intelligence (Byrne and Whiten 1989), social cognition (Aktipis 2000; Kurzban and Aktipis 2007; Kurzban and Aktipis 2008) and reasoning more generally (Sterelny 2003) may have evolved in response to social competition. Nevertheless, humans certainly have complex social abilities that can be brought to bear on social problems in ways that can promote cooperation. However, these complex abilities may not be necessary for the evolution of cooperation; cooperation may have initially evolved through much simpler means. The complexity that we see in human social life might be based on a very simple foundation: the ability to leave low-quality environments. If Walk Away is an evolutionarily ancient behavior, evolution is likely to

have created many behavioral and cognitive systems that modify it. For example, the capacities for commitment (Nesse 2001) and bonding (Carter et al. 2006) are likely candidates for systems that constrain Walk Away. In fact, bonding can be described as an unwillingness to ‘walk away’ and commitment as an inability to ‘walk away’ from social interactions.

Limitations In this paper, we argue that contingent movement promotes the evolution of cooperation without complex cognitive abilities and that the findings of the Walk Away and environmental feedback models suggest that cooperation may be the state of nature in a wider variety of contexts than previously appreciated. Here we note some limitations of contingent movement in promoting cooperation. Rules like Walk Away or other contingent movement rules are less successful when search time for new partners is very low (Enquist & Leimar 1993). When search times are low, defectors incur fewer costs from degrading their environments because they are able to easily move on to new regions. In general, the effectiveness of contingent movement in promoting cooperation relies on the implicit costs that defectors incur from searching for new groups once they’ve been abandoned by their previous group members. In general, high densities (Aktipis, 2004) make contingent movement less viable because the search time for new partners is low in these situations. However, the dyadic Walk Away model also demonstrated that very high search times can tip the balances in favor of contingent cooperation (i.e., Tit-for-Tat) rather than contingent movement (i.e., Walk Away), although defectors continue to do poorly with high search times (Aktipis, 2004). Not surprisingly, in order for contingent movement to increase the viability of cooperation, cooperators must be able to use it. If only defectors can use movement to exploit regions of cooperators, defectors are more successful (Dugatkin & Wilson 1992).

Conclusions Walk Away and environmental feedback require no complex cognitive abilities and they generate dynamics that support selection for cooperation under less restrictive assumptions than previous spatial and group selection models. The Walk Away and environmental feedback models challenge the idea that cooperation without cognition is only stable under restrictive assumptions. When individuals use contingent movement, this generates systematic migration, positive assortment, and the dissolution of uncooperative groups/aggregations. These emergent features of Walk Away all serve to increase the viability of group selection and therefore promote the evolution of cooperation.

References Cited Aktipis, C. A, and R. O Kurzban. 2004. Is homo economicus extinct? Vernon Smith, Daniel Kahneman and the evolutionary perspective. In Journal of Austrian Economics, ed. R. Koppl, 7:135-153. Amsterdam: Elsevier. Aktipis, C. A., 2000. An evolutionary perspective on consciousness: the role of emotion, theory of mind and self-deception. The Harvard Brain: Journal of Mind, Brain and Behavior 7: 29-34. Aktipis, C. A., 2004. Know when to walk away: contingent movement and the evolution of cooperation in groups. Journal of Theoretical Biology 231, no. 2: 249-260. Aktipis, C. A., 2008. When to Walk Away and when to stay: Cooperation evolves when agents can leave unproductive partners and groups. . PhD Dissertation, University of Pennsylvania. Axelrod, R. 1984. The Evolution of Cooperation. New York: Basic Books. Barclay, P, and Willer, R.. 2007. Partner choice creates competitive altruism in humans. Proceedings of the Royal Society B: Biological Sciences 274, no. 1610 (March 7): 749-753. doi:10.1098/rspb.2006.0209. Boone, R. T., and Macy, M. 1999. Unlocking the doors to prisoner's dilemma: dependence, selectivity, and cooperation. Social Psychology Quarterly 62: 32-52. Byrne, R. W., and Whiten, A. 1989. Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans (Oxford Science Publications). Oxford University Press, USA. Carter, C. S., Ahnert, L, Grossmann, K. E., Hrdy, S. B., Lamb, M. E., Porges, S. W. and Sachser N.. 2006. Attachment and Bonding: A New Synthesis. The MIT Press. Dawkins, R. 1976. The Selfish Gene. Oxford: Oxford University Press. Fletcher, J.A. and M. Doebeli. 2009. A simple and general explanation for the evolution of altruism. Proceedings of the Royal Society B-Biological Sciences 276(1654): 13-19. Hamilton, W. D. 1964a. The genetical evolution of social behavior I. J. Theor. Biol. 7: 116. Hamilton, W. D. 1964b. The geneticial evolution of social behavior II. J. Thoer. Biol. 7: 17-52. Henrich, J., et al. 2005. "Economic man" in Cross-cultural perspective: Behavioral experiments in 15 small-scale societies. Behavioral and Brain Sciences 28(6): 795-855. Hobbes, T. 1651. Leviathan. In Classics of Moral and Political Theory. Ed. Michael L. Morgan. 2nd Ed. Hackett Publishing Company. 1992. 571-732. Kurzban, R., and Aktipis, C. A.. 2007. Modularity and the Social Mind: Are Psychologists Too Self-ish? Personality and Social Psychology Review 11, no. 2 (May 1): 131-149. doi:10.1177/1088868306294906. Kurzban, R., and Aktipis, C. A.. 2008. Modular minds, multiple motives. In Evolution and Social Psychology, 39-53. Shaller, A; Simpson, J & Kenrick, D. (Eds). New York: Psychology Press. Nesse, RM., M.D. 2001. Evolution and the Capacity for Commitment. Russell Sage Foundation Publications.

Orbell, J M, Schwartz-Shea, P, and Simmons, R. T. 1984. Do cooperators exit more readily than defectors? The American Political Science Review 78, no. 1: 147-162. Pepper, J.W. & Smuts, B. B.B 2002. A Mechanism for the Evolution of Altruism among Non-Kin: Positive assortment through environmental feedback. American Naturalist 160: 205-213. Pepper, J.W. & Smuts, B.B. 2000. The evolution of cooperation in an ecological context: an agent-based model. Pp. 45-76 in: Dynamics in Human and Primate Societies: Agent-Based Modeling of Social and Spatial Processes. T. A. Kohler and G. J. Gumerman, eds. Oxford University Press, Oxford. Pepper, J.W. 2007. Simple Models of Assortment through environmental feedback. Artificial Life, 13(1): 1-9. Plato. Crito. In Classics of Moral and Political Theory. Ed. Michael L. Morgan. 2nd Ed. Hackett Publishing Company. 1992. 22-30. Rousseau, J. 1755. Discourse on the origins and foundations of inequality among men. In Classics of Moral and Political Theory. Ed. Michael L. Morgan. 2nd Ed. Hackett Publishing Company. 1992. 848-912. Skyrms, B. 2003. The Stag Hunt and the Evolution of Social Structure. Cambridge University Press. Smith, E A. 2003. Human Cooperation: Perspectives from behavioral ecology. In Genetic and cultural evolution of cooperation, ed. Peter Hammerstein, 401-427. United States of America: MIT and Freie Universität Berlin. Stephens, D.W., Brown, J.S., and Ydenberg, R.C. (eds.) 2007. Foraging: behavior and ecology. University of Chicago Press, Chicago. Sterelny, K. 2003. Thought in a Hostile World: The Evolution of Human Cognition. Wiley-Blackwell. Vanberg, V. J., and Congleton, R. D.. 1992. Rationality, morality and exit. The American Political Science Review 86, no. 2: 418-431. Waite, T.A. and Field, K.L. 2007. Foraging with others: games social foragers play. pp. 331-362 in: Stephens et al (eds.) Foraging: behavior and ecology. Chicago: University of Chicago Press. Yamagishi, T., and Hayashi, N.. 1996. Selective play: social embeddedness of social dilemmas. In Frontiers in Social Dilemma Research, ed. Wim Liebrand and David Messick, 363-384. Berlin: Springer. Yamagishi, T., Hayashi, N., and Jin, N. 1994. Prisoner's dilemma networks: Selection strategy versus action strategy. In Social dilemmas and cooperation, ed. U. Schulz, W. Albers, and U. Mueller, 233-250. Berlin: Springer-Verlag.