Indirect Reciprocity and the Evolution of Prejudicial Groups
The data here presented support a framework for the evolution of prejudicial groups, adopting indirect reciprocity as the test scenario. In particular we observe how cooperation and prejudice coevolve, with cooperation being directed in-group. This is conducted through an agent based model over a population of agents that interact through a `donation game' in which resources are donated to third parties at a cost without receiving a direct benefit. A reputation system based on social comparison is maintained to help agent decisions to donate in order to prevent exploitation by defective individuals. Two reputation values are maintained per individual representing the views of the agent’s own group (group reputation) and those of others external to the group (universal reputation). The latter reputation is diminished when individuals interact out-group by applying a prejudice factor that is characterising their own group and shared among members. The process is repeated for a number of generations in which the wealthiest individuals - where an agent’s fitness is measured by the difference between the donations received and the costs - reproduce into the next generation producing offspring that adopt their own strategy, either cooperative or uncooperative according to the relative reputation between donor and recipient.
A python script .py explaining how to run the numerical simulation is provided together with a 'readme' file that explains the setting of the variables to reproduce different types of experiments. Furthermore, some examples of the output produced by different runs in .csv format. The output produces the list of agents i with the following variables evaluated at the end of each generation:
-) the id of the agent and its trait value as immutable characteristics of the agent through generations.
-) the strategy played (s_i,u_i,d_i) in the generation representing attitudes to donate to agents with similar, higher or lower reputation than the donor.
-) the payoff accumulated at the end of the generation as the difference between benefits form donations received and costs from donations made.
-) the agent i’s prejudice value alpha_i in the discrete range {0,0.25,0.5,0.75,1}.
-) the group g_i of agent i defined uniquely by the pair (t_i,alpha_i).
-) the cooperation rate of the agent as the ratio between the donations made and requests received over the generation.
-) the reputation values at the end the generation for both group and universal reputation.
Funding
DAIS ITA BPP18 (2018-01-15 - 2020-01-14); Preece, Alun. Funder: IBM UK Ltd:4603315687
History
Specialist software required to view data files
noneLanguage(s) in dataset
- English-Great Britain (EN-GB)