The Evolution of Generosity and Choosiness in Cooperative Exchanges

✍ Scribed by Thomas N. Sherratt; Gilbert Roberts

• Publisher: Elsevier Science
• Year: 1998
• Tongue: English
• Weight: 207 KB
• Volume: 193
• Category: Article
• ISSN: 0022-5193
• DOI: 10.1006/jtbi.1998.0703

✦ Synopsis

In this paper we present a resource-explicit Donor-Receiver model for reciprocally altruistic interactions that obeys the defining inequalities of the Prisoner's Dilemma. In our model, individuals vary in the quantity of resource they invest when cooperating (termed "generosity") and they have the freedom to opt out of interactions with potential partners on the basis of their past experiences with these players (termed "choosiness"). Dynamic optimal solutions were found using a genetic algorithm in which the decision rules (cooperate or defect), generosity when cooperating, and choosiness exhibited by individuals when deciding to opt out, were all coded on genes held on two separate chromosomes. Through this genetic algorithm, individuals that had alleles which resulted in greatest success at playing our modified Prisoner's Dilemma left more offspring. When the benefit of receiving a unit resource exceeded the cost of giving, then generous cooperative behaviour tended to emerge within the population, even when the alleles of all the individuals in the starting population were set to defect. When the probability of individuals re-encountering one another was increased, individuals not only cooperated more, but they developed greater generosity. However, as the ratio of the benefits received to costs expended increased above 1, individuals in this model remained highly cooperative but their median generosity decreased significantly. In contrast to earlier studies using genetic algorithms, the extra potential for cheating afforded by asymmetrical degrees of generosity meant that genuinely cooperative behaviour did not emerge in the equivalent round-robin tournament in which individuals were not able to exercise partner preference.