Exploiting Objects as Artifacts in Multi-Agent Based Social Simulations

In this study a recent evolution and learning model for artifacts is extended to address the ability of artificial social agents to realize their goals by adapting the exploitation of dynamic artifacts in dynamic environments over time. An implemented case study is provided incorporating the model into the multi-agent simulation of the Village EcoDynamics Project developed to study the early Pueblo Indian settlers from A.D. 600 to 1300. The dynamic landscape used for settling and farming is abstracted as an artifact and agents learn to adapt its exploitation over time by employing individual, social and population learning strategies. Comparing various strategies revealed learning through social networks while evolving the extent of the network as the best adaptive strategy. The results are consistent with archeological records as a wider margin is observed between social and non-social learners during periods known for the highest landscape variability. In addition, learning through social networks outperforms learning via cultural beliefs which is expected given the heterogeneity of the landscape.