Developing an Agent-based Simulation of the Propagation of False Beliefs

Context/Background Using multi-agent systems to create social simulations is a new and exciting methodology, allowing scientists to study and predict the emergence of social phenomena. In 1998, a simulation was produced by Jim Doran to study the consequences of false beliefs spread among a population. His findings suggested that in some situations, false beliefs can in fact be beneficial to a society. Aims This study aims to reproduce Doran’s model and obtain similar results. A second model is also designed to investigate how social bonds within a society, in this case friendship, can affect the spread of false beliefs and the impact of those beliefs on society. Method The simulation toolkit RePast is used as a development framework to build the models. In the first model, agents exist in an environment that contains food and a ‘deathzone’. Agents cannot perceive the deathzone, but any agent that enters will immediately die. Agents compete with each other for food. Some agents believe in ‘pseudoagents’, that is, they see agents where none exist. Agents may communicate beliefs with each other. The second model will be similar, but in this case agents will only communicate beliefs with agents they consider ‘friends’. In both models I will be observing the beliefs held by the population and the survivability of a population under a given set of parameters. Results Collective misbelief in the agent society had a significant impact of the environment and growth of the population, but in no cases did it produce an observable benefit to the agents. The number of food resources in the environment is significantly higher when agents hold false beliefs. Conclusions The findings of this experiment disagree with the results reported by Jim Doran in his paper, but there is plenty of scope for further experimentation with these models, such as investigating the impact of agent misbelief on other elements in the environment.