A Simulation of Temporally Variant Agent Interaction via Belief Promulgation

Message Driven Communication

Thus far, we have focused on the exploration of the globally emergent behaviors in passive agent interaction systems. The agents reacted to their environment, but did so in a manner where each agent’s vivification was independent of neighboring vivifications. In the message based version of this simulation, the focus shifts from agents behaving passively within the environment into a model where each agent actively attempts to exert influence over the environment. The emergent behaviors observed in previous sections resulted from agents examining their immediate surroundings and updating themselves accordingly. Global behavior arose from the non-deterministic agent vivification order and the asynchronous nature of the updates. In this set of experiments, global emergence is driven by the exchange of messages.

In this section, we expanded our simulation to accommodate active agents which directly communicate—albeit in a primitive manner. Information is exchanged as simple messages which are reflective of an agent’s internal state. Though agents may take on many states during a simulation, each agent communicates its active state with its spatially embedded neighbors. The active model is divided into two distinct subtypes. The first subtype, discussed in Section “Message Driven Game of Life”, is a direct extension of the previous “Conway” model; but agents respond to events generated by neighbors rather than vivificating autonomously. The second subtype, discussed in Section “Fuzzy ‘Belief’ Promulgation”, is a completely new model based upon temporally variant “belief” interaction. The models in both subtypes display interesting and rather unique behavioral characteristics.