Solidarity, synchronization and the emergence of cooperation arXiv ...
arXiv:1312.6809v16 [physics.soc-ph] 12 Jul 2015
Solidarity, synchronization and the emergence of cooperation Jeroen Bruggeman∗ July 14, 2015
Abstract One of the big questions about social life is how people manage to cooperate for public goods. Current answers rely on individuals’ relations and reputations, but have difficulty explaining the onset, when relations are sparse and reputations largely unknown. The challenge rises when realizing that actual situations are often characterized by uncertainties about the value and timing of the public good, as well as the consequences of contributions. As a run up to cooperation, people can establish social ties and increase their solidarity through interaction rituals. Kuramoto’s synchronization model is used to show that the network thereby formed should have sufficient algebraic connectivity to compensate for the differences between individuals’ commitments and psychological states. At a critical level of solidarity, a majority’s commitments and psychological states, respectively, synchronize in a phase transition, which yields a boost of motivation for a burst of collective action, even under multifold uncertainty.
A remarkable feature of humans is that they can act collectively—and effectively at that—while individuals are tempted to defect and exploit the results of others’ efforts [1]. In contrast to current models and lab experiments, many of those situations are characterized by multi-fold uncertainty. For example, at defending settlements, mass protests and revolts against dictatorial regimes, contributing may be retaliated by the opponent, costs can ∗
Department of Sociology, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV Amsterdam, the Netherlands. Email: [email protected]. Thank you to Randall Collins, Jacquelien van Stekelenburg and Justus Uitermark for comments.
1
be unexpectedly high, participants may be paralysed by fear [2], and benefits may differ from those expected or arrive later, if at all. Yet history shows that even under adverse conditions, people often cooperate [3]. Whether individuals know their costs and benefits or guess inaccurately under uncertainty, their decisions to contribute certainly depend on what they know about others. It’s crucial for them to form a group wherein members are related, which in general can be by network ties, a high chance of future encounters, kinship, proximity, shared foci, or a combination thereof [1, 4]. Relations provide access to reputations [5, 6, 7], which summarize others’ cooperativeness [8], but in a broader sense also encompass other’s interests, skills and behavior toward free riders. Reputations make possible to choose, reward, avoid, punish or ostracize specific individuals [6, 9, 10], thus feed back into the relations [7], and are to be backed up by cooperative norms to work properly [9, 11]. At the beginning, however, when relations are sparse and reputations largely unknown, the challenge is to get cooperation off the ground. Individuals have to establish (or re-enforce) social relations first and, in the absence of reputations, develop a shared intentionality [12] with respect to a given collective good: the same level of commitment to it, the same thoughts and emotions about their endeavour, and an awareness of this sharedness. Here we pose the question how people can achieve this under adverse conditions. Under favourable conditions, with a low cost-to-benefit ratio and uncertainty only about the number of contributors, the onset of cooperation can be explained more easily, as discussed later.
Result To start out, members of a (becoming) group with a shared focus can perform certain interaction rituals [13, 14], a subset of all rituals exercised by humans. Examples are marching and noise making at street demonstrations before physical confrontation with the incumbent power [15]; dance and religious ceremonies before a hunt or fight [14, 16]; drill to prepare for combat [16]; and, team building activities in organizations to enhance workplace performance [17]. These preparatory actions come at relatively low cost compared to contributions to the pertaining public good later on, and may sometimes suffice without further actions to achieve a low-cost political goal. Many rituals involve rhythmic entrainment, which generates feedback that reinforces perceptions of similarity [18]. In experiments comparing treat2
ments with asynchronous to synchronous movements, contributions to public goods were significantly higher in the latter [18, 19, 20]. Interaction rituals increase solidarity [13, 14, 21], also called identification [22, 23] or loyalty [24, 25], which denote the bonding strength of individuals to a group. Moral justification, religion or sacred values [19, 26] can enhance emotional intensity that further increases solidarity [13, 16]. The higher the uncertainties, the stronger group-directed emotions should be aroused. A competitive group [27] or an enemy [2, 28] can give an extra push. Rare but emotionally intense rituals, e.g. initiation in the French Foreign Legion, have a stronger effect on solidarity than frequently occurring low arousal rituals such as prayers [29, 30]. At the start of an interaction ritual, N participants will usually have different commitments to the public good in question. Commitments are modelled as a symmetric single peaked distribution g(ω) mean-centred at 0, for example a Gaussian. During the ritual, these commitments may increase along with solidarity, sometimes under committed leadership [31], but high average commitments do not predict cooperation [32]. Lowly committed may be tempted to free ride on the highly committed, who in turn may distrust the former and withdraw. Cooperation is higher among equally committed, as was shown experimentally [33]. For participants to learn about each other’s commitments, as well as to sense each others’ body language and emotions, they should interact in physical co-presence [13, 34], a key feature of interaction rituals. This makes possible for commitments to become (local) common knowledge [35]. These interactions are modelled as symmetric ties aij = aji = 1 (and absent ties aij = 0) among participants indexed i and j, later generalized to aij ≥ 0. Participants have fluctuating thoughts and emotions about their undertaking, modelled as psychological states θi (t). The change of i’s state is affected by i’s commitment ωi and, through empathy [36], by the psychological states of i’s social contacts [37]. The effect size of the latter is determined by their difference, θj (t)−θi (t), multiplied by solidarity; at higher solidarity, the influence of other group members is stronger. For notational clarity, time indices are now dropped. To analyse the effect of interaction rituals, Kuramoto’s well-studied model [38, 39, 40] is used. It has already solved numerous problems in physics, biology, engineering, complex networks and computer science [41, 42, 43], thereby establishing cross-disciplinary parsimony. By definition, the smaller a stable difference between θj and θi , the higher their degree of synchronization [44], 3
which in our case can be made visible by hyperscanning [45]. In Kuramoto’s original model, ω is a frequency and θ a phase. Obviously, people are no oscillators, but we can get tractability through these simplifications. The reading of coupling strength K as solidarity is straightforward. When writing θ˙i as a shorthand for dθi (t)/dt, N KX θ˙i = ωi + aij sin(θj − θi ). N j=1
(1)
The degree to which all group members are synchronized is indicated by an order parameter 0 ≤ r ≤ 1 (to be precise, a complex parameter r(t)eiφ(t) where φ is the average phase [40]). Analytic solutions were derived for complete graphs, wherein every node is connected to every other node and N is very large. When solidarity increases, nothing happens initially and psychological states remain incoherent (r = 0). At a critical threshold Kc , however, there is a sudden transition toward stable, although not perfect (r < 1), synchronization of a large majority, and Eq.1 implies that commitments synchronize in the same moment. This two-fold phase transition becomes “explosive” when taking into account that solidarity varies across individuals and is correlated with commitments, and Ki = K|ωi | is substituted for K in Eq.(1) [46]. A phase transition toward synchronization has also been found for many sparse graphs with finite N [41]. Social networks, except for very small groups, are sparse, clustered into subgroups, have skewed degree distributions (numbers of social contacts), and short network distances [4, 47, 48]. Solidarity is limited by the nervous system, and can’t reach arbitrarily high values. For a social network to synchronize at feasible solidarity, its connectivity has to compensate for the differences between individuals’ commitments, else r jitters and synchronization is not achieved [44, 42]. Stability of r is studied by means of the Laplacian matrix of the graph [49]. If in its spectrum, 0 = λ1 ≤ λ2 ≤ · · · ≤ λN , there are m (near) zero eigenvalues, they indicate the presence of m (almost) disconnected graph components that won’t synchronize with each other. It turns out that increasing a network’s algebraic connectivity [50]—λ2 —yields synchronization at lower solidarity [41, 42]. Density, average distance [41] and degree distribution [51] do not predict this in general. To illustrate, Fig.1 compares a “wheel” (λ2 = 2) to a “bow tie” (λ2 = 1) topology, which are equal in size (7), density (0.57), average distance (1.43), 4
degree distribution, degree centralization (0.6), and coreness [52] (both are 3cores). For each draw of initial values (from a uniform distribution between −π and π) and commitments (from a Gaussian with σ = 2), the wheel synchronizes at a lower Kc than the bow tie. Notice that in small graphs, such as these, very small differences between these values have a large impact on Kc . Algebraic connectivity can increase by increasing tie strengths, but people have limited capability to do so, at the expense of other ties [53]. This social homeostasis [54] also constrains the number of ties that individuals can maintain. Members connecting by strong ties (aij 1) can only do this if their group is small and for a limited time. In Fig.1, bow tie’s connectivity would double by a two-fold increase of all tie strengths, whereas relaying two ties to create a wheel has the same effect. Topological modifications with a given number of ties (and keeping aij = 1) are clearly more efficient. Alternatively, connectivity can increase by increasing group size, but as large groups inevitably cluster into subgroups [48], there is a—yet unknown— maximum connectivity. Well-connected subgroups s synchronize at Ks < Kc [55], and if there is assortment, or homophily [56], of commitments such that their variation ||ωs || is relatively small, synchronization happens at lower solidarity, net of connectivity [44, 42]. Remainder group members may then be an audience that supports those subgroups [2]. In sum, if participant’s interaction ritual is in full swing and their algebraic connectivity is sufficient, the model predicts that their psychological states and commitments, respectively, fuse into one. The simultaneity of synchronization will yield a stronger boost of motivation, experienced as “collective effervescence” [14], than if it were sequential, and collective protests, for example, are bursty indeed [3]. Moreover, experiments show that when acting spontaneously, people are more cooperative than when they calculate their decisions [57]. At some point during or after a (series of) collective action(s), participants or their resources will be exhausted and their solidarity will decrease [13]. If commitments are correlated with solidarity [46] or with degree [58], there is hysteresis: the backward transition from synchrony to asynchrony happens at lower solidarity than the forward transition, and initial differences are recovered. After a successful collective action, when also a network with reputations and norms is established, future actions by the same group will be easier to mount, and can be often explained by familiar mechanisms of cooperation. 5
1.0
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
0.0
0.2
0.4
r
0.6
0.8
●
0
1
2
3
4
5
K
Figure 1: Wheel and bow tie graphs, and their synchronization r as a function of solidarity K, indicated by triangles and dots, respectively. In large graphs at K < Kc , r = 0 but in these small graphs r jitters with time (not shown) and there is no synchronization by definition. Depending on the initial values, here 20 draws, synchronization starts at a solidarity value between the dashed lines.
Discussion Cooperation under adverse conditions has a cold start problem, which can be solved by interaction rituals. When a group gets to collective action, it might look as if it’s members are set off by a spark [63]—a minor event such as a police arrest of some protesters. The model, however, explains sparks’ effect in terms of a small increase of solidarity in a near-critical group, due to an emotional reaction that would not entail collective action well below the critical threshold. Interaction rituals also provide a solution to the start up problem of critical mass theory [59, 60, 61], which has it that if a sufficient number of people contributes first, they will win over the rest. Others attempt to explain the onset of cooperation in terms of individuals’ identities collapsing into one group identity [32, 62]. If identity fusion is interpreted as achieving a shared intentionality, our model provides an explanation for their findings. When uncertainties and expected costs are relatively low, a mellow ritual will do or is not even necessary. People can then chat with each other about the public good [64], thereby establishing social ties and exchanging 6
commitments [65], thoughts and emotions. As in interaction rituals, face to face contact is essential [64, 66], but high solidarity is not, and synchronization can be loosened to (approximate) consensus. Yet algebraic connectivity is important to achieve consensus fast, and for information transmission to be reliable, explained below. In a graph colouring experiment [67], algebraic connectivity turned out to be inversely proportional to the time to reach consensus. There, subjects had to choose the same colour as their network-neighbours, in highly clustered networks and in networks with randomly rewired edges, in other words, in networks with progressively higher algebraic connectivity [68]. In cooperation experiments without communication, network topologies were unimportant for the level of contributions [69, 70]. In those experiments, subjects could respond to others’ actions in previous rounds by either cooperating or defecting with everybody in their neighbourhood [69] or group [70]. For topology to have an effect, however, participants must be able to reciprocate the (in)actions of specific others [71], else free riding is inconsequential for some whereas others are punished for cooperating. This coarse grained behavior is clearly inefficient, and for fine grained reciprocity with specific individuals, and for the diffusion of reputations, topology does matter. For the diffusion of information under realistic conditions, social networks should be robust against noise—misinterpreted, mistransmitted or manipulated information—and against node removal [72]. These requirements motivated a definition of social cohesion as the minimum number, κ, of independent paths (concatenation of ties) connecting arbitrary pairs of nodes in a network [73]. This number is equivalent to the minimum number of nodes that has to be removed to make the network fall apart [74]. Only in very small networks, of, say, 7 members of a team, everyone can be connected directly to everyone else (in this example, κ = 6 and λ2 = 7). It was proven for all incomplete networks (missing at least one tie) that λ2 ≤ κ [50, 75]. Algebraic connectivity thus not only indicates synchronization potential in a broad sense, including consensus and robustness against small perturbations of psychological states [49], but also a lower bound for social cohesion and redundancy of information channels. It might even be important for other species. Orca’s, for example, can collectively catch a seal that sits on a slab of floating ice by swimming synchronously to create a wave that washes it off [76], hence they must exchange information in their network to pull this off. Low algebraic connectivity, in contrast, facilitates anti-coordination, e.g. choosing a different color for oneself than one’s neighbours in a graph 7
coloring game [67], which is useful when group members want to differentiate themselves to exchange private goods instead of achieving public goods. To study dynamic networks, the Kuramoto model has been generalized as follows [54]. Ties aij (t) ≥ 0 strengthen between people who perceive each other as similar in terms of their θ’s, which happens under the constraint of homeostasis. A tie that disappears is modeled as a fading tie, aij (t) → 0. Starting out with a random network wherein solidarity increases from zero to a low value, subgroups emerge that are internally synchronous but mutually asynchronous. If solidarity continues to increase to K ≥ Kc , those subgroups merge into one synchronized group [77]. When taking into account that solidarity varies across individuals, represented by a Gaussian distribution of Ki , it takes longer for synchronous subgroups to emerge, and individuals with very low Ki stay solitary, out of sync with everyone else [78]. These loners set apart, the overall pattern is qualitatively the same as with one K for all. In this study on the onset of cooperation, Kuramoto’s model was used to predict that groups with higher algebraic connectivity will start faster, all else being equal. Under adverse conditions, the onset was explained by interaction rituals leading to synchronization. The latter is predicted to happen in a phase transition that shows up in a burst of cooperation.
Methods The synchronization model is elaborated in refs [40, 41, 44, 46, 42], and dynamic networks in refs [54, 77].
References [1] Rand, D. A. & Nowak, M. A. Human cooperation. Trends in Cognitive Science 18, 413–425 (2013). [2] Collins, R. Violence: A Micro-Sociological Theory (Princeton University Press, Princeton, 2008). [3] Tarrow, S. Power in Movement: Social Movements, Collective Action and Politics (Cambridge University Press, Cambridge, 1994). [4] Apicella, C. L., Marlowe, F. W., Fowler, J. H. & Christakis, N. A. Social networks and cooperation in hunter-gatherers. Nature 481, 497–501 (2012).
8
[5] Nowak, M. A. & Sigmund, K. Evolution of indirect reciprocity. Nature 437, 1291–1298 (2005). [6] Cuesta, J. A., Gracia-L´ azaro, C., Ferrer, A., Moreno, Y. & S´anchez, A. Reputation drives cooperative behaviour and network formation in human groups. Scientific Reports 5 (2015). [7] Gallo, E. & Yan, C. The effects of reputational and social knowledge on cooperation. Proceedings of the National Academy of Sciences 112, 3647– 3652 (2015). [8] Peysakhovich, A., Nowak, M. A. & Rand, D. G. Humans display a cooperative phenotype that is domain general and temporally stable. Nature Communications 5, 4939 (2014). [9] Seinen, I. & Schram, A. Social status and group norms: Indirect reciprocity in a repeated helping experiment. European Economic Review 50, 581–602 (2006). [10] Olson, M. The Logic of Collective Action: Public Goods and the Theory of Groups (Harvard University Press, Harvard, Mass, 1965). [11] Fehr, E. & Fischbacher, U. Social norms and human cooperation. Trends in Cognitive Sciences 8, 185–190 (2004). [12] Tomasello, M., Carpenter, M., Call, J., Behne, T. & Moll, H. Understanding and sharing intentions: The origins of cultural cognition. Behavioral and Brain Sciences 28, 675–691 (2005). [13] Collins, R. Interaction Ritual Chains (Princeton University Press, Princeton, 2004). [14] Durkheim, E. Les formes ´el´ementaires de la vie religieuse (Presses Universitaires de France, Paris, 1912). [15] McPhail, C. & Wohlstein, R. T. Collective locomotion as collective behavior. American Sociological Review 447–463 (1986). [16] McNeill, W. H. Keeping Together in Time: Dance and Drill in Human History (Harvard University Press, Cambridge, Mass, 1995). [17] Kunda, G. Engineering Culture (Temple, Philadelphia, 1992). [18] Reddish, P., Fischer, R. & Bulbulia, J. Lets dance together: Synchrony, shared intentionality and cooperation. PLoS ONE 8, e71182 (2013).
9
[19] Fischer, R., Callander, R., Reddish, P. & Bulbulia, J. How do rituals affect cooperation? Human Nature 24, 115–125 (2013). [20] Wiltermuth, S. S. & Heath, C. Synchrony and cooperation. Psychological Science 20, 1–5 (2009). [21] Whitehouse, H. & Lanman, J. A. The ties that bind us. Current Anthropology 55, 674–695 (2014). [22] Klandermans, B. How group identification helps to overcome the dilemma of collective action. American Behavioral Scientist 45, 887–900 (2002). [23] Kelly, C. Group identification, intergroup perceptions and collective action. European Review of Social Psychology 4, 59–83 (1993). [24] Simon, H. A. A mechanism for social selection and successful altruism. Science 250, 1665–1668 (1990). [25] Diamond, J. Economics: The wealth of nations. Nature 429, 616–617 (2004). [26] Atran, S. & Ginges, J. Religious and sacred imperatives in human conflict. Science 336, 855–857 (2012). [27] Sherif, M. et al. Intergroup Conflict and Cooperation: The Robbers Cave Experiment (Institute of Group Relations, University of Oklahoma, Norman, OK, 1961). [28] Choi, J.-K. & Bowles, S. The coevolution of parochial altruism and war. Science 318, 636–640 (2007). [29] Atkinson, Q. D. & Whitehouse, H. The cultural morphospace of ritual form: Examining modes of religiosity cross-culturally. Evolution and Human Behavior 32, 50–62 (2011). [30] Jones, D. The ritual animal. Nature 493, 470–472 (2013). [31] Kori, H. & Mikhailov, A. S. Entrainment of randomly coupled oscillator networks by a pacemaker. Physical Review Letters 93, 254101 (2004). [32] Swann Jr, W. B., Jetten, J., G´omez, A., Whitehouse, H. & Brock, B. When group membership gets personal: A theory of identity fusion. Psychological Review 119, 441 (2012). [33] Kurzban, R., McCabe, K., Smith, V. L. & Wilson, B. J. Incremental commitment and reciprocity in a real-time public goods game. Personality and Social Psychology Bulletin 27, 1662–1673 (2001).
10
[34] Sebanz, N., Bekkering, H. & Knoblich, G. Joint action: bodies and minds moving together. Trends in Cognitive Sciences 10, 70–76 (2006). [35] Chwe, M. S.-Y. Rational Ritual: Culture, Coordination, and Common Knowledge (Princeton University Press, Princeton, 2001). [36] De Waal, F. B. The antiquity of empathy. Science 59, 874–876 (2012). [37] Konvalinka, I. et al. Synchronized arousal between performers and related spectators in a fire-walking ritual. Proceedings of the National Academy of Sciences 108, 8514–8519 (2011). [38] Nadis, S. All together now. Nature 421, 780–782 (2003). [39] Kuramoto, Y. Self-entrainment of a population of coupled non-linear oscillators. In International Symposium on Mathematical Problems in Theoretical Physics, vol. 39, 420–422 (1975). [40] Strogatz, S. H. From Kuramoto to Crawford: exploring the onset of synchronization in populations of coupled oscillators. Physica D 143, 1–20 (2000). [41] Arenas, A., D´ıaz-Guilerac, A., Kurths, J., Moreno, Y. & Zhou, C. Synchronization in complex networks. Physics Reports 469, 93–153 (2008). [42] D¨ orfler, F. & Bullo, F. Synchronization in complex networks of phase oscillators: A survey. Automatica 50, 1539–1564 (2014). [43] Kocarev, L. (ed.) Consensus and Synchronization in Complex Networks (Springer, Heidelberg, 2013). [44] Jadbabaie, A., Motee, N. & Barahona, M. On the stability of the Kuramoto model of coupled nonlinear oscillators. In Proceedings of the American Control Conference, 4296–4301 (2004). [45] Bilek, E. et al. Information flow between interacting human brains: Identification, validation, and relationship to social expertise. Proceedings of the National Academy of Sciences 112, 5207–5212 (2015). [46] Hu, X. et al. Exact solution for first-order synchronization transition in a generalized Kuramoto model. Scientific Reports 4, 7262 (2014). [47] Newman, M. E. The structure and function of complex networks. SIAM Review 45, 167–256 (2003).
11
[48] Garlaschelli, D., Ahnert, S. E., Fink, T. M. A. & Caldarelli, G. Lowtemperature behaviour of social and economic networks. Entropy 15, 3148– 3169 (2013). [49] McGraw, P. N. & Menzinger, M. Analysis of nonlinear synchronization dynamics of oscillator networks by Laplacian spectral methods. Physical Review E 75, 027104 (2007). [50] Fiedler, M. Algebraic connectivity of graphs. Czechoslovak Mathematical Journal 23, 298–305 (1973). [51] Atay, F. M., Biyikoglu, T. & Jost, J. Synchronization of networks with prescribed degree distributions. IEEE Transactions on Circuits and Systems I 53, 92–98 (2006). [52] Seidman, S. B. Network structure and minimum degree. Social Networks 5, 269–287 (1983). [53] Miritello, G. et al. Time as a limited resource: Communication strategy in mobile phone networks. Social Networks 35, 89–95 (2013). [54] Assenza, S., Guti´errez, R., G´omez-Garde˜ nes, J. & Boccaletti, S. Emergence of structural patterns out of synchronization in networks with competitive interactions. Scientific Reports 1, 1–8 (2011). [55] Arenas, A., D´ıaz-Guilera, A. & P´erez-Vicente, C. J. Synchronization reveals topological scales in complex networks. Physical Review Letters 96, 114102 (2006). [56] McPherson, J. M., Smith-Lovin, L. & Cook, J. M. Birds of a feather: Homophily in social networks. Annual Review of Sociology 27, 415–444 (2001). [57] Rand, D. G., Greene, J. D. & Nowak, M. A. Spontaneous giving and calculated greed. Nature 489, 427–430 (2012). [58] G´ omez-Garde˜ nes, J., G´ omez, S., Arenas, A. & Moreno, Y. Explosive synchronization transitions in scale-free networks. Physical Review Letters 106, 128701 (2011). [59] Schelling, T. C. Micromotives and Macrobehavior (W.W. Norton, New York, 1978). [60] Granovetter, M. Threshold models of collective behavior. American Journal of Sociology 83, 1420–1443 (1978).
12
[61] Marwell, G. & Oliver, P. The Critical Mass in Collective Action (Cambridge University Press, Cambridge, MA, 1993). [62] Whitehouse, H., McQuinn, B., Buhrmester, M. & Swann Jr., W. B. Brothers in arms: Libyan revolutionaries bond like family. Proceedings of the National Academy of Sciences 111, 17783–17785 (2014). [63] Zedong, M. A single spark can start a prairie fire (Foreign Languages Press, Michigan, 1953). [64] Sally, D. Conversation and cooperation in social dilemmas a meta-analysis of experiments from 1958 to 1992. Rationality and Society 7, 58–92 (1995). [65] Ostrom, E. A behavioral approach to the rational choice theory of collective action. American Political Science Review 92, 1–22 (1998). [66] De Rooij, E. A., Green, D. P. & Gerber, A. S. Field experiments on political behavior and collective action. Annual Review of Political Science 12, 389– 395 (2009). [67] Judd, S., Kearns, M. & Vorobeychik, Y. Behavioral dynamics and influence in networked coloring and consensus. Proceedings of the National Academy of Sciences 107, 14978–14982 (2010). [68] Olfati-Saber, R. Ultrafast consensus in small-world networks. In IEEE Proceedings of the American Control Conference, 2371–2378 (2005). [69] Gracia-L´ azaro, C. et al. Heterogeneous networks do not promote cooperation when humans play a prisoners dilemma. Proceedings of the National Academy of Sciences 109, 12922–12926 (2012). [70] Suri, S. & Watts, D. J. Cooperation and contagion in web-based, networked public goods experiments. PLoS ONE 6, e16836 (2011). [71] Panchanathan, K. & Boyd, R. Indirect reciprocity can stabilize cooperation without the second-order free rider problem. Nature 432, 499–502 (2004). [72] Moody, J. & White, D. R. Structural cohesion and embeddedness: A hierarchical concept of social groups. American Sociological Review 68, 103–127 (2003). [73] White, D. R. & Harary, F. The cohesiveness of blocks in social networks: Node connectivity and conditional density. Sociological Methodology 31, 305–359 (2001).
13
[74] Harary, F. Graph Theory (Perseus, Reading, MA, 1969). [75] de Abreu, N. M. M. Old and new results on algebraic connectivity of graphs. Linear Algebra and its Applications 423, 53–73 (2007). [76] Visser, I. N. et al. Antarctic peninsula killer whales (orcinus orca) hunt seals and a penguin on floating ice. Marine Mammal Science 24, 225–234 (2008). [77] Guti´errez, R. et al. Emerging meso- and macroscales from synchronization of adaptive networks. Physical Review Letters 107, 234103 (2011). [78] Bruggeman, J. & P´eli, G. Small firm subsistence and market dimensionality. International Journal of Modern Physics C 25, 1450053 (2014).
14
Solidarity, synchronization and the emergence of cooperation Jeroen Bruggeman∗ July 14, 2015
Abstract One of the big questions about social life is how people manage to cooperate for public goods. Current answers rely on individuals’ relations and reputations, but have difficulty explaining the onset, when relations are sparse and reputations largely unknown. The challenge rises when realizing that actual situations are often characterized by uncertainties about the value and timing of the public good, as well as the consequences of contributions. As a run up to cooperation, people can establish social ties and increase their solidarity through interaction rituals. Kuramoto’s synchronization model is used to show that the network thereby formed should have sufficient algebraic connectivity to compensate for the differences between individuals’ commitments and psychological states. At a critical level of solidarity, a majority’s commitments and psychological states, respectively, synchronize in a phase transition, which yields a boost of motivation for a burst of collective action, even under multifold uncertainty.
A remarkable feature of humans is that they can act collectively—and effectively at that—while individuals are tempted to defect and exploit the results of others’ efforts [1]. In contrast to current models and lab experiments, many of those situations are characterized by multi-fold uncertainty. For example, at defending settlements, mass protests and revolts against dictatorial regimes, contributing may be retaliated by the opponent, costs can ∗
Department of Sociology, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV Amsterdam, the Netherlands. Email: [email protected]. Thank you to Randall Collins, Jacquelien van Stekelenburg and Justus Uitermark for comments.
1
be unexpectedly high, participants may be paralysed by fear [2], and benefits may differ from those expected or arrive later, if at all. Yet history shows that even under adverse conditions, people often cooperate [3]. Whether individuals know their costs and benefits or guess inaccurately under uncertainty, their decisions to contribute certainly depend on what they know about others. It’s crucial for them to form a group wherein members are related, which in general can be by network ties, a high chance of future encounters, kinship, proximity, shared foci, or a combination thereof [1, 4]. Relations provide access to reputations [5, 6, 7], which summarize others’ cooperativeness [8], but in a broader sense also encompass other’s interests, skills and behavior toward free riders. Reputations make possible to choose, reward, avoid, punish or ostracize specific individuals [6, 9, 10], thus feed back into the relations [7], and are to be backed up by cooperative norms to work properly [9, 11]. At the beginning, however, when relations are sparse and reputations largely unknown, the challenge is to get cooperation off the ground. Individuals have to establish (or re-enforce) social relations first and, in the absence of reputations, develop a shared intentionality [12] with respect to a given collective good: the same level of commitment to it, the same thoughts and emotions about their endeavour, and an awareness of this sharedness. Here we pose the question how people can achieve this under adverse conditions. Under favourable conditions, with a low cost-to-benefit ratio and uncertainty only about the number of contributors, the onset of cooperation can be explained more easily, as discussed later.
Result To start out, members of a (becoming) group with a shared focus can perform certain interaction rituals [13, 14], a subset of all rituals exercised by humans. Examples are marching and noise making at street demonstrations before physical confrontation with the incumbent power [15]; dance and religious ceremonies before a hunt or fight [14, 16]; drill to prepare for combat [16]; and, team building activities in organizations to enhance workplace performance [17]. These preparatory actions come at relatively low cost compared to contributions to the pertaining public good later on, and may sometimes suffice without further actions to achieve a low-cost political goal. Many rituals involve rhythmic entrainment, which generates feedback that reinforces perceptions of similarity [18]. In experiments comparing treat2
ments with asynchronous to synchronous movements, contributions to public goods were significantly higher in the latter [18, 19, 20]. Interaction rituals increase solidarity [13, 14, 21], also called identification [22, 23] or loyalty [24, 25], which denote the bonding strength of individuals to a group. Moral justification, religion or sacred values [19, 26] can enhance emotional intensity that further increases solidarity [13, 16]. The higher the uncertainties, the stronger group-directed emotions should be aroused. A competitive group [27] or an enemy [2, 28] can give an extra push. Rare but emotionally intense rituals, e.g. initiation in the French Foreign Legion, have a stronger effect on solidarity than frequently occurring low arousal rituals such as prayers [29, 30]. At the start of an interaction ritual, N participants will usually have different commitments to the public good in question. Commitments are modelled as a symmetric single peaked distribution g(ω) mean-centred at 0, for example a Gaussian. During the ritual, these commitments may increase along with solidarity, sometimes under committed leadership [31], but high average commitments do not predict cooperation [32]. Lowly committed may be tempted to free ride on the highly committed, who in turn may distrust the former and withdraw. Cooperation is higher among equally committed, as was shown experimentally [33]. For participants to learn about each other’s commitments, as well as to sense each others’ body language and emotions, they should interact in physical co-presence [13, 34], a key feature of interaction rituals. This makes possible for commitments to become (local) common knowledge [35]. These interactions are modelled as symmetric ties aij = aji = 1 (and absent ties aij = 0) among participants indexed i and j, later generalized to aij ≥ 0. Participants have fluctuating thoughts and emotions about their undertaking, modelled as psychological states θi (t). The change of i’s state is affected by i’s commitment ωi and, through empathy [36], by the psychological states of i’s social contacts [37]. The effect size of the latter is determined by their difference, θj (t)−θi (t), multiplied by solidarity; at higher solidarity, the influence of other group members is stronger. For notational clarity, time indices are now dropped. To analyse the effect of interaction rituals, Kuramoto’s well-studied model [38, 39, 40] is used. It has already solved numerous problems in physics, biology, engineering, complex networks and computer science [41, 42, 43], thereby establishing cross-disciplinary parsimony. By definition, the smaller a stable difference between θj and θi , the higher their degree of synchronization [44], 3
which in our case can be made visible by hyperscanning [45]. In Kuramoto’s original model, ω is a frequency and θ a phase. Obviously, people are no oscillators, but we can get tractability through these simplifications. The reading of coupling strength K as solidarity is straightforward. When writing θ˙i as a shorthand for dθi (t)/dt, N KX θ˙i = ωi + aij sin(θj − θi ). N j=1
(1)
The degree to which all group members are synchronized is indicated by an order parameter 0 ≤ r ≤ 1 (to be precise, a complex parameter r(t)eiφ(t) where φ is the average phase [40]). Analytic solutions were derived for complete graphs, wherein every node is connected to every other node and N is very large. When solidarity increases, nothing happens initially and psychological states remain incoherent (r = 0). At a critical threshold Kc , however, there is a sudden transition toward stable, although not perfect (r < 1), synchronization of a large majority, and Eq.1 implies that commitments synchronize in the same moment. This two-fold phase transition becomes “explosive” when taking into account that solidarity varies across individuals and is correlated with commitments, and Ki = K|ωi | is substituted for K in Eq.(1) [46]. A phase transition toward synchronization has also been found for many sparse graphs with finite N [41]. Social networks, except for very small groups, are sparse, clustered into subgroups, have skewed degree distributions (numbers of social contacts), and short network distances [4, 47, 48]. Solidarity is limited by the nervous system, and can’t reach arbitrarily high values. For a social network to synchronize at feasible solidarity, its connectivity has to compensate for the differences between individuals’ commitments, else r jitters and synchronization is not achieved [44, 42]. Stability of r is studied by means of the Laplacian matrix of the graph [49]. If in its spectrum, 0 = λ1 ≤ λ2 ≤ · · · ≤ λN , there are m (near) zero eigenvalues, they indicate the presence of m (almost) disconnected graph components that won’t synchronize with each other. It turns out that increasing a network’s algebraic connectivity [50]—λ2 —yields synchronization at lower solidarity [41, 42]. Density, average distance [41] and degree distribution [51] do not predict this in general. To illustrate, Fig.1 compares a “wheel” (λ2 = 2) to a “bow tie” (λ2 = 1) topology, which are equal in size (7), density (0.57), average distance (1.43), 4
degree distribution, degree centralization (0.6), and coreness [52] (both are 3cores). For each draw of initial values (from a uniform distribution between −π and π) and commitments (from a Gaussian with σ = 2), the wheel synchronizes at a lower Kc than the bow tie. Notice that in small graphs, such as these, very small differences between these values have a large impact on Kc . Algebraic connectivity can increase by increasing tie strengths, but people have limited capability to do so, at the expense of other ties [53]. This social homeostasis [54] also constrains the number of ties that individuals can maintain. Members connecting by strong ties (aij 1) can only do this if their group is small and for a limited time. In Fig.1, bow tie’s connectivity would double by a two-fold increase of all tie strengths, whereas relaying two ties to create a wheel has the same effect. Topological modifications with a given number of ties (and keeping aij = 1) are clearly more efficient. Alternatively, connectivity can increase by increasing group size, but as large groups inevitably cluster into subgroups [48], there is a—yet unknown— maximum connectivity. Well-connected subgroups s synchronize at Ks < Kc [55], and if there is assortment, or homophily [56], of commitments such that their variation ||ωs || is relatively small, synchronization happens at lower solidarity, net of connectivity [44, 42]. Remainder group members may then be an audience that supports those subgroups [2]. In sum, if participant’s interaction ritual is in full swing and their algebraic connectivity is sufficient, the model predicts that their psychological states and commitments, respectively, fuse into one. The simultaneity of synchronization will yield a stronger boost of motivation, experienced as “collective effervescence” [14], than if it were sequential, and collective protests, for example, are bursty indeed [3]. Moreover, experiments show that when acting spontaneously, people are more cooperative than when they calculate their decisions [57]. At some point during or after a (series of) collective action(s), participants or their resources will be exhausted and their solidarity will decrease [13]. If commitments are correlated with solidarity [46] or with degree [58], there is hysteresis: the backward transition from synchrony to asynchrony happens at lower solidarity than the forward transition, and initial differences are recovered. After a successful collective action, when also a network with reputations and norms is established, future actions by the same group will be easier to mount, and can be often explained by familiar mechanisms of cooperation. 5
1.0
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
0.0
0.2
0.4
r
0.6
0.8
●
0
1
2
3
4
5
K
Figure 1: Wheel and bow tie graphs, and their synchronization r as a function of solidarity K, indicated by triangles and dots, respectively. In large graphs at K < Kc , r = 0 but in these small graphs r jitters with time (not shown) and there is no synchronization by definition. Depending on the initial values, here 20 draws, synchronization starts at a solidarity value between the dashed lines.
Discussion Cooperation under adverse conditions has a cold start problem, which can be solved by interaction rituals. When a group gets to collective action, it might look as if it’s members are set off by a spark [63]—a minor event such as a police arrest of some protesters. The model, however, explains sparks’ effect in terms of a small increase of solidarity in a near-critical group, due to an emotional reaction that would not entail collective action well below the critical threshold. Interaction rituals also provide a solution to the start up problem of critical mass theory [59, 60, 61], which has it that if a sufficient number of people contributes first, they will win over the rest. Others attempt to explain the onset of cooperation in terms of individuals’ identities collapsing into one group identity [32, 62]. If identity fusion is interpreted as achieving a shared intentionality, our model provides an explanation for their findings. When uncertainties and expected costs are relatively low, a mellow ritual will do or is not even necessary. People can then chat with each other about the public good [64], thereby establishing social ties and exchanging 6
commitments [65], thoughts and emotions. As in interaction rituals, face to face contact is essential [64, 66], but high solidarity is not, and synchronization can be loosened to (approximate) consensus. Yet algebraic connectivity is important to achieve consensus fast, and for information transmission to be reliable, explained below. In a graph colouring experiment [67], algebraic connectivity turned out to be inversely proportional to the time to reach consensus. There, subjects had to choose the same colour as their network-neighbours, in highly clustered networks and in networks with randomly rewired edges, in other words, in networks with progressively higher algebraic connectivity [68]. In cooperation experiments without communication, network topologies were unimportant for the level of contributions [69, 70]. In those experiments, subjects could respond to others’ actions in previous rounds by either cooperating or defecting with everybody in their neighbourhood [69] or group [70]. For topology to have an effect, however, participants must be able to reciprocate the (in)actions of specific others [71], else free riding is inconsequential for some whereas others are punished for cooperating. This coarse grained behavior is clearly inefficient, and for fine grained reciprocity with specific individuals, and for the diffusion of reputations, topology does matter. For the diffusion of information under realistic conditions, social networks should be robust against noise—misinterpreted, mistransmitted or manipulated information—and against node removal [72]. These requirements motivated a definition of social cohesion as the minimum number, κ, of independent paths (concatenation of ties) connecting arbitrary pairs of nodes in a network [73]. This number is equivalent to the minimum number of nodes that has to be removed to make the network fall apart [74]. Only in very small networks, of, say, 7 members of a team, everyone can be connected directly to everyone else (in this example, κ = 6 and λ2 = 7). It was proven for all incomplete networks (missing at least one tie) that λ2 ≤ κ [50, 75]. Algebraic connectivity thus not only indicates synchronization potential in a broad sense, including consensus and robustness against small perturbations of psychological states [49], but also a lower bound for social cohesion and redundancy of information channels. It might even be important for other species. Orca’s, for example, can collectively catch a seal that sits on a slab of floating ice by swimming synchronously to create a wave that washes it off [76], hence they must exchange information in their network to pull this off. Low algebraic connectivity, in contrast, facilitates anti-coordination, e.g. choosing a different color for oneself than one’s neighbours in a graph 7
coloring game [67], which is useful when group members want to differentiate themselves to exchange private goods instead of achieving public goods. To study dynamic networks, the Kuramoto model has been generalized as follows [54]. Ties aij (t) ≥ 0 strengthen between people who perceive each other as similar in terms of their θ’s, which happens under the constraint of homeostasis. A tie that disappears is modeled as a fading tie, aij (t) → 0. Starting out with a random network wherein solidarity increases from zero to a low value, subgroups emerge that are internally synchronous but mutually asynchronous. If solidarity continues to increase to K ≥ Kc , those subgroups merge into one synchronized group [77]. When taking into account that solidarity varies across individuals, represented by a Gaussian distribution of Ki , it takes longer for synchronous subgroups to emerge, and individuals with very low Ki stay solitary, out of sync with everyone else [78]. These loners set apart, the overall pattern is qualitatively the same as with one K for all. In this study on the onset of cooperation, Kuramoto’s model was used to predict that groups with higher algebraic connectivity will start faster, all else being equal. Under adverse conditions, the onset was explained by interaction rituals leading to synchronization. The latter is predicted to happen in a phase transition that shows up in a burst of cooperation.
Methods The synchronization model is elaborated in refs [40, 41, 44, 46, 42], and dynamic networks in refs [54, 77].
References [1] Rand, D. A. & Nowak, M. A. Human cooperation. Trends in Cognitive Science 18, 413–425 (2013). [2] Collins, R. Violence: A Micro-Sociological Theory (Princeton University Press, Princeton, 2008). [3] Tarrow, S. Power in Movement: Social Movements, Collective Action and Politics (Cambridge University Press, Cambridge, 1994). [4] Apicella, C. L., Marlowe, F. W., Fowler, J. H. & Christakis, N. A. Social networks and cooperation in hunter-gatherers. Nature 481, 497–501 (2012).
8
[5] Nowak, M. A. & Sigmund, K. Evolution of indirect reciprocity. Nature 437, 1291–1298 (2005). [6] Cuesta, J. A., Gracia-L´ azaro, C., Ferrer, A., Moreno, Y. & S´anchez, A. Reputation drives cooperative behaviour and network formation in human groups. Scientific Reports 5 (2015). [7] Gallo, E. & Yan, C. The effects of reputational and social knowledge on cooperation. Proceedings of the National Academy of Sciences 112, 3647– 3652 (2015). [8] Peysakhovich, A., Nowak, M. A. & Rand, D. G. Humans display a cooperative phenotype that is domain general and temporally stable. Nature Communications 5, 4939 (2014). [9] Seinen, I. & Schram, A. Social status and group norms: Indirect reciprocity in a repeated helping experiment. European Economic Review 50, 581–602 (2006). [10] Olson, M. The Logic of Collective Action: Public Goods and the Theory of Groups (Harvard University Press, Harvard, Mass, 1965). [11] Fehr, E. & Fischbacher, U. Social norms and human cooperation. Trends in Cognitive Sciences 8, 185–190 (2004). [12] Tomasello, M., Carpenter, M., Call, J., Behne, T. & Moll, H. Understanding and sharing intentions: The origins of cultural cognition. Behavioral and Brain Sciences 28, 675–691 (2005). [13] Collins, R. Interaction Ritual Chains (Princeton University Press, Princeton, 2004). [14] Durkheim, E. Les formes ´el´ementaires de la vie religieuse (Presses Universitaires de France, Paris, 1912). [15] McPhail, C. & Wohlstein, R. T. Collective locomotion as collective behavior. American Sociological Review 447–463 (1986). [16] McNeill, W. H. Keeping Together in Time: Dance and Drill in Human History (Harvard University Press, Cambridge, Mass, 1995). [17] Kunda, G. Engineering Culture (Temple, Philadelphia, 1992). [18] Reddish, P., Fischer, R. & Bulbulia, J. Lets dance together: Synchrony, shared intentionality and cooperation. PLoS ONE 8, e71182 (2013).
9
[19] Fischer, R., Callander, R., Reddish, P. & Bulbulia, J. How do rituals affect cooperation? Human Nature 24, 115–125 (2013). [20] Wiltermuth, S. S. & Heath, C. Synchrony and cooperation. Psychological Science 20, 1–5 (2009). [21] Whitehouse, H. & Lanman, J. A. The ties that bind us. Current Anthropology 55, 674–695 (2014). [22] Klandermans, B. How group identification helps to overcome the dilemma of collective action. American Behavioral Scientist 45, 887–900 (2002). [23] Kelly, C. Group identification, intergroup perceptions and collective action. European Review of Social Psychology 4, 59–83 (1993). [24] Simon, H. A. A mechanism for social selection and successful altruism. Science 250, 1665–1668 (1990). [25] Diamond, J. Economics: The wealth of nations. Nature 429, 616–617 (2004). [26] Atran, S. & Ginges, J. Religious and sacred imperatives in human conflict. Science 336, 855–857 (2012). [27] Sherif, M. et al. Intergroup Conflict and Cooperation: The Robbers Cave Experiment (Institute of Group Relations, University of Oklahoma, Norman, OK, 1961). [28] Choi, J.-K. & Bowles, S. The coevolution of parochial altruism and war. Science 318, 636–640 (2007). [29] Atkinson, Q. D. & Whitehouse, H. The cultural morphospace of ritual form: Examining modes of religiosity cross-culturally. Evolution and Human Behavior 32, 50–62 (2011). [30] Jones, D. The ritual animal. Nature 493, 470–472 (2013). [31] Kori, H. & Mikhailov, A. S. Entrainment of randomly coupled oscillator networks by a pacemaker. Physical Review Letters 93, 254101 (2004). [32] Swann Jr, W. B., Jetten, J., G´omez, A., Whitehouse, H. & Brock, B. When group membership gets personal: A theory of identity fusion. Psychological Review 119, 441 (2012). [33] Kurzban, R., McCabe, K., Smith, V. L. & Wilson, B. J. Incremental commitment and reciprocity in a real-time public goods game. Personality and Social Psychology Bulletin 27, 1662–1673 (2001).
10
[34] Sebanz, N., Bekkering, H. & Knoblich, G. Joint action: bodies and minds moving together. Trends in Cognitive Sciences 10, 70–76 (2006). [35] Chwe, M. S.-Y. Rational Ritual: Culture, Coordination, and Common Knowledge (Princeton University Press, Princeton, 2001). [36] De Waal, F. B. The antiquity of empathy. Science 59, 874–876 (2012). [37] Konvalinka, I. et al. Synchronized arousal between performers and related spectators in a fire-walking ritual. Proceedings of the National Academy of Sciences 108, 8514–8519 (2011). [38] Nadis, S. All together now. Nature 421, 780–782 (2003). [39] Kuramoto, Y. Self-entrainment of a population of coupled non-linear oscillators. In International Symposium on Mathematical Problems in Theoretical Physics, vol. 39, 420–422 (1975). [40] Strogatz, S. H. From Kuramoto to Crawford: exploring the onset of synchronization in populations of coupled oscillators. Physica D 143, 1–20 (2000). [41] Arenas, A., D´ıaz-Guilerac, A., Kurths, J., Moreno, Y. & Zhou, C. Synchronization in complex networks. Physics Reports 469, 93–153 (2008). [42] D¨ orfler, F. & Bullo, F. Synchronization in complex networks of phase oscillators: A survey. Automatica 50, 1539–1564 (2014). [43] Kocarev, L. (ed.) Consensus and Synchronization in Complex Networks (Springer, Heidelberg, 2013). [44] Jadbabaie, A., Motee, N. & Barahona, M. On the stability of the Kuramoto model of coupled nonlinear oscillators. In Proceedings of the American Control Conference, 4296–4301 (2004). [45] Bilek, E. et al. Information flow between interacting human brains: Identification, validation, and relationship to social expertise. Proceedings of the National Academy of Sciences 112, 5207–5212 (2015). [46] Hu, X. et al. Exact solution for first-order synchronization transition in a generalized Kuramoto model. Scientific Reports 4, 7262 (2014). [47] Newman, M. E. The structure and function of complex networks. SIAM Review 45, 167–256 (2003).
11
[48] Garlaschelli, D., Ahnert, S. E., Fink, T. M. A. & Caldarelli, G. Lowtemperature behaviour of social and economic networks. Entropy 15, 3148– 3169 (2013). [49] McGraw, P. N. & Menzinger, M. Analysis of nonlinear synchronization dynamics of oscillator networks by Laplacian spectral methods. Physical Review E 75, 027104 (2007). [50] Fiedler, M. Algebraic connectivity of graphs. Czechoslovak Mathematical Journal 23, 298–305 (1973). [51] Atay, F. M., Biyikoglu, T. & Jost, J. Synchronization of networks with prescribed degree distributions. IEEE Transactions on Circuits and Systems I 53, 92–98 (2006). [52] Seidman, S. B. Network structure and minimum degree. Social Networks 5, 269–287 (1983). [53] Miritello, G. et al. Time as a limited resource: Communication strategy in mobile phone networks. Social Networks 35, 89–95 (2013). [54] Assenza, S., Guti´errez, R., G´omez-Garde˜ nes, J. & Boccaletti, S. Emergence of structural patterns out of synchronization in networks with competitive interactions. Scientific Reports 1, 1–8 (2011). [55] Arenas, A., D´ıaz-Guilera, A. & P´erez-Vicente, C. J. Synchronization reveals topological scales in complex networks. Physical Review Letters 96, 114102 (2006). [56] McPherson, J. M., Smith-Lovin, L. & Cook, J. M. Birds of a feather: Homophily in social networks. Annual Review of Sociology 27, 415–444 (2001). [57] Rand, D. G., Greene, J. D. & Nowak, M. A. Spontaneous giving and calculated greed. Nature 489, 427–430 (2012). [58] G´ omez-Garde˜ nes, J., G´ omez, S., Arenas, A. & Moreno, Y. Explosive synchronization transitions in scale-free networks. Physical Review Letters 106, 128701 (2011). [59] Schelling, T. C. Micromotives and Macrobehavior (W.W. Norton, New York, 1978). [60] Granovetter, M. Threshold models of collective behavior. American Journal of Sociology 83, 1420–1443 (1978).
12
[61] Marwell, G. & Oliver, P. The Critical Mass in Collective Action (Cambridge University Press, Cambridge, MA, 1993). [62] Whitehouse, H., McQuinn, B., Buhrmester, M. & Swann Jr., W. B. Brothers in arms: Libyan revolutionaries bond like family. Proceedings of the National Academy of Sciences 111, 17783–17785 (2014). [63] Zedong, M. A single spark can start a prairie fire (Foreign Languages Press, Michigan, 1953). [64] Sally, D. Conversation and cooperation in social dilemmas a meta-analysis of experiments from 1958 to 1992. Rationality and Society 7, 58–92 (1995). [65] Ostrom, E. A behavioral approach to the rational choice theory of collective action. American Political Science Review 92, 1–22 (1998). [66] De Rooij, E. A., Green, D. P. & Gerber, A. S. Field experiments on political behavior and collective action. Annual Review of Political Science 12, 389– 395 (2009). [67] Judd, S., Kearns, M. & Vorobeychik, Y. Behavioral dynamics and influence in networked coloring and consensus. Proceedings of the National Academy of Sciences 107, 14978–14982 (2010). [68] Olfati-Saber, R. Ultrafast consensus in small-world networks. In IEEE Proceedings of the American Control Conference, 2371–2378 (2005). [69] Gracia-L´ azaro, C. et al. Heterogeneous networks do not promote cooperation when humans play a prisoners dilemma. Proceedings of the National Academy of Sciences 109, 12922–12926 (2012). [70] Suri, S. & Watts, D. J. Cooperation and contagion in web-based, networked public goods experiments. PLoS ONE 6, e16836 (2011). [71] Panchanathan, K. & Boyd, R. Indirect reciprocity can stabilize cooperation without the second-order free rider problem. Nature 432, 499–502 (2004). [72] Moody, J. & White, D. R. Structural cohesion and embeddedness: A hierarchical concept of social groups. American Sociological Review 68, 103–127 (2003). [73] White, D. R. & Harary, F. The cohesiveness of blocks in social networks: Node connectivity and conditional density. Sociological Methodology 31, 305–359 (2001).
13
[74] Harary, F. Graph Theory (Perseus, Reading, MA, 1969). [75] de Abreu, N. M. M. Old and new results on algebraic connectivity of graphs. Linear Algebra and its Applications 423, 53–73 (2007). [76] Visser, I. N. et al. Antarctic peninsula killer whales (orcinus orca) hunt seals and a penguin on floating ice. Marine Mammal Science 24, 225–234 (2008). [77] Guti´errez, R. et al. Emerging meso- and macroscales from synchronization of adaptive networks. Physical Review Letters 107, 234103 (2011). [78] Bruggeman, J. & P´eli, G. Small firm subsistence and market dimensionality. International Journal of Modern Physics C 25, 1450053 (2014).
14
