Abstract
This paper investigates an evolutionary iterated prisoner’s dilemma (IPD) model of multiple agents, in which agents interact in terms of the pair-wise IPD game while adapting their attitudes towards income stream risk. Specifically, agents will become more risk averse (or more risk seeking) if their game payoffs exceed (or fall below) their expectations. In particular, agents use their peers’ average payoffs as expectations (social comparison) when their payoffs are lower than their peers’ averages, but use their own historical payoffs as expectations (historical comparison) when their payoffs are higher than their peers’ averages. Such selective attention to social comparison or historical comparison manifests a desire for continuous improvement of agents. Simulations are conducted to investigate the evolution of cooperation under the selective attention mechanism. Results indicate that agents can sustain a highly cooperative equilibrium when they consider selective attention in adjusting their risk attitudes. This holds true for both the well-mixed and the network-based games, even in the presence of uncertain game payoffs. The reason is that, selective attention can significantly induce agents to adhere to conditional cooperation as well as to identify uncertainty in payoffs, which enhances the risk-averse behavior of agents in the IPD game. As a result, high levels of cooperation can be attained.
Similar content being viewed by others
Notes
The prospect theory used the terms expectations and aspirations interchangeably (Kahneman and Tversky 1979).
Here, what we mean is that, individuals or organizations tend to use their own past performances as expectations (i.e., historical comparison) if they outperform their peers’ averages, but they tend to use their peers’ average performances as expectations (i.e., social comparison) if they underperform their peers’ averages. It is worth noting that the prerequisite for the application of historical comparison incorporates a social-comparison-like process, in which individuals or organizations need to confirm that they do be above their peers’ averages before implementing historical comparison. However, for the purpose of clarity, we explicitly term this circumstance as “historical comparison”, because it is the individuals’ or the organizations’ own past performances that are directly used as their expectations in this circumstance. In other words, it is historical comparison that mainly determines the individuals’ or the organizations’ expectations. On the other hand, when the term “social comparison” is presented, it refers to the case in which individuals or organizations directly use their peers’ average performances as their expectations.
Agents with \(\alpha > 0.8\) are found to perform similarly well in the IPD game.
The literature has suggested that an IPD strategy should cover histories of only recent interactions, because only recent moves have significance for the current move (Darwen and Yao 1997). The consideration of previous three moves has been widely exploited in the literature (Axelrod 1984; Franken and Engelbrecht 2005; Mittal and Deb 2009).
The game-playing process is independent across generations, which means that each agent will start new IPD games with \(g_{n}\) new opponents, respectively, in a new generation. Still, each IPD game entails l encounters.
Note that the reference-performance selection process is cost-effective, because each agent uses only information on its own payoff in the prior generation and information on the population average payoff in the current generation. The population average payoff, like, the average profit of an industry or a sub-industry, is always public information.
According to previous work in Zeng et al. (2016a, 2017), the value of \(\gamma\) indicates the sensitivity of changes in agents’ risk attitudes. A too-small value of \(\gamma\), e.g., \(\gamma = 0\) (or a too-large value of \(\gamma\), e.g., \(\gamma = 0.5\)), means that the agents might change their risk attitudes too frequently (or too slowly) in response to the game outcomes. These inappropriate adjustment speeds may prevent agents from adapting to the game environment, and, therefore, have a disruptive impact on the evolutionary outcome. Thus, the value of \(\gamma\) is set as 0.15 in this study, as suggested in Zeng et al. (2016a). Meanwhile, because comparable results are obtained for different specifications of \(r_{up}\) and \(r_{down}\), we use \(r_{up} = r_{down} = 0.2\) for illustration. These parameter settings are also convenient for the comparison of results with selective attention (i.e., results in the current study) to those without (Zeng et al. 2016a, 2017).
Simulation results are qualitatively identical with the population size and the average neighborhood size in a wide range of values, respectively. We specify N as 256 for convenience in constructing the grid network as a \(16 \times 16\) lattice. In addition, the neighborhood is defined as the \(5 \times 5\) lattice around a focal agent in the grid network, which results in \(24 = 5 \times 5 - 1\) neighbors for each agent. The construction of the other networks is relatively easy with any given population size and neighborhood size. Furthermore, the specification of population size as 256 and the definition of neighborhood size as 24 help in providing comparable results to those achieved in our previous studies (Zeng et al. 2016a, 2017), in which homogeneous historical comparison or social comparison was used for agents’ risk attitude adaptation. Note that locality of interaction can be guaranteed when the neighborhood size of 24 is compared to the population size of 256, which brings heterogeneity into interactions of agents (Zeng et al. 2017).
Note that \(\xi_{1} {, }\xi_{{2}} {, }\xi_{{3}} {, }\xi_{{4}}\) are independently drawn for the four payoffs \(T,R,P,S\). The resulting \((T^{\prime},R^{\prime},P^{\prime},S^{\prime})\) that do not satisfy the constraint \(T^{\prime} > R^{\prime} > P^{\prime} > S^{\prime}\) or \(2R^{\prime} > T^{\prime} + S^{\prime}\) will be excluded for sustaining the IPD-payoff pattern.
In our experiments, the average result over 20 runs is not statistically different from the average results over 30, 40, or 50 runs (with a 95% confidence level).
The value of 2.6 is used here as an indication of the emergence of high levels of cooperation, which is higher than the average payoff of 2.5 that agents obtain from the move “Cooperate–Defect” or “Defect–-Cooperate” (Fogel 1995).
The clustering coefficients of the ring, ring-based small-world, scale-free, grid, grid-based small-world, and random networks are 0.72, 0.63, 0.75, 0.52, 0.46, and 0.09, respectively; and their characteristic path lengths are 5.82, 2.58, 2.14, 2.93, 2.46, and 2.00, respectively. According to Jun and Sethi (2007), the clustering coefficient indicates the degree to which agents form local clusters, and the characteristic path length implies the average distance between agents in the network.
Experiments using smaller neighborhood sizes or larger population sizes, which generally involve longer characteristic path lengths, also indicate that long characteristic path lengths impede global cooperation in the network-based IPD game.
Although the population average risk attitude declines from about 0.9 to about 0.6 as \(\sigma\) is changed from 0 to 3 (Fig. 8a), agents with risk attitudes of around 0.6, similar to those with 0.9, can consistently cooperate with each other in the IPD game (Zeng et al. 2016b). In addition, an inspection on results by generation shows that the population average risk attitude persistently increases to and finally stabilizes around 0.6 when \(\sigma = 3\), which indicates the steady formation of global cooperation in the evolution.
References
Aranda C, Arellano J, Davila A (2017) Organizational learning in target setting. Acad Manag J 60(3):1189–1211
Axelrod R (1984) The evolution of cooperation. Basic Books, New York
Baggio JA, Papyrakis E (2014) Agent-based simulations of subjective well-being. Soc Indic Res 115(2):623–635
Blettner DP, He ZL, Hu S, Bettis RA (2015) Adaptive aspirations and performance heterogeneity: attention allocation among multiple reference points. Strateg Manag J 36(7):987–1005
Boyd R, Lorberbaum JP (1987) No pure strategy is evolutionarily stable in the repeated prisoner's dilemma game. Nature 327(6117):58–59
Bromiley P (1991) Testing a causal model of corporate risk taking and performance. Acad Manag J 34(1):37–59
Chen S-H (2008) Software-agent designs in economics: an interdisciplinary. IEEE Comput Intell Mag 3(4):18–22
Chen S-H (2012) Varieties of agents in agent-based computational economics: a historical and an interdisciplinary perspective. J Econ Dyn Control 36(1):1–25
Chen Y-S, Yang H-X, Guo W-Z (2017) Aspiration-induced dormancy promotes cooperation in the spatial prisoner’s dilemma games. Physica A 469:625–630
Chiong R, Kirley M (2012) Effects of iterated interactions in multiplayer spatial evolutionary games. IEEE Trans Evol Comput 16(4):537–555
Cyert RM, March JG (1963) A behavioral theory of the firm. Prentice Hall, Englewood Cliffs, NJ
Darwen PJ, Yao X (1997) Speciation as automatic categorical modularization. IEEE Trans Evol Comput 1(2):101–108
Dato S, Grunewald A, Müller D, Strack P (2017) Expectation-based loss aversion and strategic interaction. Games Econ Behav 104(Supplement C):681–705
Daylamani-Zad D, Agius H, Angelides MC (2020) Reflective agents for personalisation in collaborative games. Artif Intell Rev 53:429–474
Dreber A, Fudenberg D, Rand DG (2014) Who cooperates in repeated games: the role of altruism, inequity aversion, and demographics. J Econ Behav Organ 98:41–55
Fafchamps M, Kebede B, Zizzo DJ (2015) Keep up with the winners: experimental evidence on risk taking, asset integration, and peer effects. Eur Econ Rev 79:59–79
Festinger L (1954) A theory of social comparison processes. Hum Relat 7(2):117–140
Fiegenbaum A (1990) Prospect theory and the risk-return association: an empirical examination in 85 industries. J Econ Behav Organ 14(2):187–203
Fiegenbaum A, Thomas H (1995) Strategic groups as reference groups: theory, modeling and empirical examination of industry and competitive strategy. Strateg Manag J 16(6):461–476
Fogel DB (1995) On the relationship between the duration of an encounter and the evolution of cooperation in the iterated prisoner's dilemma. Evol Comput 3(3):349–363
Franken N, Engelbrecht AP (2005) Particle swarm optimization approaches to coevolve strategies for the iterated prisoner's dilemma. IEEE Trans Evol Comput 9(6):562–579
Fudenberg D, Imhof LA (2008) Monotone imitation dynamics in large populations. J Econ Theory 140(1):229–245
Fudenberg D, Maskin E (1986) The folk theorem in repeated games with discounting or with incomplete information. Econometrica 54(3):533–554
García J, van Veelen M (2016) In and out of equilibrium I: evolution of strategies in repeated games with discounting. J Econ Theory 161:161–189
García J, van Veelen M (2018) No strategy can win in the repeated prisoner's dilemma: linking game theory and computer simulations. Front Robot AI 5:102
Ghatak A, Mukherjee D, Mallikarjuna Rao KS (2018) A spatial game theoretic analysis of conflict and identity. Comput Econ 52(2):493–519
Hodgson GM, Huang K (2012) Evolutionary game theory and evolutionary economics: are they different species? J Evol Econ 22(2):345–366
Holton GA (2004) Defining risk. Financ Anal J 60(6):19–25
Howley E, O’Riordan C (2006) The effects of viscosity in choice and refusal IPD environments. Artif Intell Rev 26(1):103–114
Ioannou C (2014) Coevolution of finite automata with errors. J Evol Econ 24(3):541–571
Ishibuchi H, Ohyanagi H, Nojima Y (2011) Evolution of strategies with different representation schemes in a spatial iterated prisoner's dilemma game. IEEE Trans Comput Intell AI Games 3(1):67–82
Joseph J, Gaba V (2015) The fog of feedback: ambiguity and firm responses to multiple aspiration levels. Strateg Manag J 36(13):1960–1978
Jun T, Sethi R (2007) Neighborhood structure and the evolution of cooperation. J Evol Econ 17(5):623–646
Jun T, Sethi R (2009) Reciprocity in evolving social networks. J Evol Econ 19(3):379–396
Kahneman D, Tversky A (1979) Prospect theory: an analysis of decision under risk. Econom J Econom Soc 47:263–291
Klemm K, Eguiluz VM (2002) Growing scale-free networks with small-world behavior. Phys Rev E 65(5):057102
Li J, Zhang C, Sun Q, Chen Z, Zhang J (2017) Changing the intensity of interaction based on individual behavior in the iterated prisoner's dilemma game. IEEE Trans Evol Comput 21(4):506–517
March JG, Shapira Z (1987) Managerial perspectives on risk and risk taking. Manag Sci 33(11):1404–1418
Miller KD, Bromiley P (1990) Strategic risk and corporate performance: an analysis of alternative risk measures. Acad Manag J 33(4):756–779
Mittal S, Deb K (2009) Optimal strategies of the iterated prisoner's dilemma problem for multiple conflicting objectives. IEEE Trans Evol Comput 13(3):554–565
Moliterno TP, Beck N, Beckman CM, Meyer M (2014) Knowing your place: social performance feedback in good times and bad times. Organ Sci 25(6):1684–1702
Muto N (2014) Strategic complexity in repeated extensive games. Games Econ Behav 83:45–52
Nowak MA, Sasaki A, Taylor C, Fudenberg D (2004) Emergence of cooperation and evolutionary stability in finite populations. Nature 428(6983):646–650
Osório A (2018) Brownian signals: information quality, quantity and timing in repeated games. Comput Econ 52(2):387–404
Park KM (2007) Antecedents of convergence and divergence in strategic positioning: the effects of performance and aspiration on the direction of strategic change. Organ Sci 18(3):386–402
Płatkowski T (2015) Aspiration-based full cooperation in finite systems of players. Appl Math Comput 251:46–54
Rong Z-H, Zhao Q, Wu Z-X, Zhou T, Tse CK (2016) Proper aspiration level promotes generous behavior in the spatial prisoner’s dilemma game. Eur Phys J B 89(7):166
Safarzyńska K, van den Bergh JC (2010) Evolutionary models in economics: a survey of methods and building blocks. J Evol Econ 20(3):329–373
Shutters ST (2012) Punishment leads to cooperative behavior in structured societies. Evol Comput 20(2):301–319
Stahl DO (2011) Cooperation in the sporadically repeated prisoners’ dilemma via reputation mechanisms. J Evol Econ 21(4):687–702
Stahl DO (2013) An experimental test of the efficacy of a simple reputation mechanism to solve social dilemmas. J Econ Behav Organ 94:116–124
Sujil A, Verma J, Kumar R (2018) Multi agent system: concepts, platforms and applications in power systems. Artif Intell Rev 49(2):153–182
Thaler RH (2005) Advances in behavioral finance. Princeton University Press, Princeton
Tomasello M (2000) The cultural origins of human cognition. Harvard University Press, Cambridge, MA
Uzzi B, Amaral LAN, Reed-Tsochas F (2007) Small-world networks and management science research: a review. Eur Manag Rev 4(2):77–91
van Veelen M, García J, Rand DG, Nowak MA (2012) Direct reciprocity in structured populations. Proc Natl Acad Sci 109(25):9929–9934
Wang H, Yan J, Yu J (2017a) Reference-dependent preferences and the risk–return trade-off. J Financ Econ 123(2):395–414
Wang IK, Qian L, Lehrer M (2017b) From technology race to technology marathon: a behavioral explanation of technology advancement. Eur Manag J 35(2):187–197
Watts DJ, Strogatz SH (1998) Collective dynamics of 'small-world' networks. Nature 393(6684):440–442
Wilson AJ, Wu H (2017) At-will relationships: how an option to walk away affects cooperation and efficiency. Games Econ Behav 102(Supplement C):487–507
Xian Y, Chen P (2011) Does social welfare preference always promote cooperation on Barabási and Albert networks? Comput Econ 37(3):249–266
Xie N-G, Zhen K-X, Wang C, Ye Y, Wang L (2015) Evolution of cooperation driven by the diversity of emotions. Connect Sci 27(1):89–101
Zeng W, Li M, Chen F (2016a) Cooperation in the evolutionary iterated prisoner’s dilemma game with risk attitude adaptation. Appl Soft Comput 44(Supplement C):238–254
Zeng W, Li M, Chen F, Nan G (2016b) Risk consideration and cooperation in the iterated prisoner’s dilemma. Soft Comput 20(2):567–587
Zeng W, Li M, Feng N (2017) The effects of heterogeneous interaction and risk attitude adaptation on the evolution of cooperation. J Evol Econ 27(3):435–459
Zhang H (2018) Errors can increase cooperation in finite populations. Games Econ Behav 107(Supplement C):203–219
Zhang H, Gao M, Wang W, Liu Z (2014) Evolutionary prisoner׳s dilemma game on graphs and social networks with external constraint. J Theor Biol 358:122–131
Zschache J (2012) Producing public goods in networks: some effects of social comparison and endogenous network change. Soc Netw 34(4):539–548
Acknowledgements
This study was supported by National Natural Science Foundation of China (Grant No. 71702040), and Natural Science Foundation of Hainan Province, China (Grant No. 20167244). The authors would also like to thank the High Performance Computing Centre (HPCC) of Tianjin University for providing the computing support.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Zeng, W., Li, M. Selective attention to historical comparison or social comparison in the evolutionary iterated prisoner’s dilemma game. Artif Intell Rev 53, 6043–6078 (2020). https://doi.org/10.1007/s10462-020-09842-5
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10462-020-09842-5
Keywords
- Iterated prisoner’s dilemma game
- Evolution of cooperation
- Selective attention
- Historical comparison
- Social comparison
- Risk attitude adaptation