Coupled MMASS: A Formal Model for Non-deterministic Multi-agent Simulations

The Multilayered Multi-agent Situated System (MMASS) has been proposed as a general framework to build multiagent systems in which agents are situated in an environment whose characterization can be multifaceted (each facet of the environment is named a layer, hence the name of this framework). Agents in the MMASS can be purely reactive and computationally lightweight, or deliberative and employ highly sophisticated reasoning mechanisms. As a consequence, the MMASS has proven to be useful to build massively multi-agent systems (in which typically each agent is computationally simple) as well as systems comprised by complex agents (in which typically we have few agents interacting with each other and with the environment). In the present article we combine a simplified version of MMASS with a specific logical system, which we suggest that can be particularly suitable to solve problems based on simulations. The proposed logical system is a variation of classical FOPL, in which logical statements are tagged with probability values so that one can reason probabilistically as opposed of reasoning with probabilities or reasoning about probabilities.