Towards supporting psychologically plausible variability in agent-based human modelling

We describe the initial steps in developing an agentbased cognitive architecture designed to support psychologically plausible human variability. The new architecture, COJACK, is based on JACK, a BDI-based agent language. It will constrain the agents to reason and act in a psychologically plausible manner. Their information processing will be adjusted by a set of parameters that moderate the agent´s reasoning and actions, combined with a set of guidelines for developing plans and beliefs for the agents. This set of parameters will also support varying the agents´ performance, both in terms of differences across agents as well as differences that arise within an individual due to internal and external factors. We conclude that other architectures will want to include a similar set, including representing a body, its interaction with the environment, and the passage of time.