[C17] Evolution and Genetics & Interdisciplinary Studies

[C17-04] The evolution of cooperation when players privately evaluate others' reputations based on personal experience and gossip with various lies

*Yosuke Fukuda¹, Mayuko Nakamaru¹ (1. Institute of Science Tokyo (Japan))

Keywords:

Evolutionary Game Theory,Gossip,Reputation,Lie,Indirect Reciprocity

Indirect reciprocity promotes the evolution of cooperation through a system of reputations, even in the absence of direct interactions between individuals. Prior studies have shown that indirect reciprocity fosters cooperation via a public assessment system in which individuals share a common reputation or via a private assessment system in which reputational information is exchanged through pairwise communication; in these systems, third parties observe players’ actions in games and update the reputations of both the actor and the recipient based on the action and their existing reputations. However, players do not always use these assessment systems to evaluate others' reputation; players may evaluate others based on their own experiences in the game and the gossip received from others. This study focuses on the latter case. Unlike the former case, which only considers the spread of false reputation created by errors, the latter case can model false gossip, including intentional lies that spread between individuals. This study investigates a mathematical model of private assessment, where reputation is shaped by two sources: personal experience in the donation game and gossip, whose reliability may be uncertain. There are 3,456 possible types of strategies, and we focus on three types of defectors and three types of conditional cooperators. Among 3,456 possible strategies, we focus on three defector types and three conditional cooperator types. The defectors are: GBBG-Defectors, who lie perversely about others (labeling good as bad and vice versa) and claim to be good themselves; GGBG-Defectors, who say everyone is good and also claim to be good; and H-Defectors, who tell the truth about others and claim to be bad. The conditional cooperators are: H-DISC, who tells the truth, says “I am good,” and prioritizes gossip; S-DISC, who tells the truth, says nothing about themselves, and prioritizes personal experience; and TFT_LIKE, who uses only personal experience. Through analysis and simulation, we find that gossip plays a crucial role in shaping trust and evolutionary dynamics, even when it includes lies. Without lies, H-DISC can outperform H-Defectors. However, when lies from GBBG-Defectors and GGBG-Defectors are present, H-DISC can no longer win, but S-DISC, which ignores unreliable gossip, can prevail. Our findings indicate that conditional cooperators can evolve when (i) experience-based information rather than gossip-based information is used to evaluate others' reputation, (ii) players evaluate the credibility of information sources and selectively accept them, and (iii) players only refer to the latest gossip. A strategy incorporating (i)-(iii) is S-DISC, which promotes the evolution of cooperation even in the parameter region where TFT_LIKE cannot win defectors except specific types of defectors.