Social simulation often concerns the behaviour of humans interacting within some system. Simulation applications are increasingly requiring more realistic and complex human modelling, than reactive rules. We suggest that the established Belief Desire Intention (BDI) approach to modelling cognitive agents, can usefully be applied to modelling humans in social simulations. Traditional social science resources can be used to develop models of human decision making and behaviour that can be represented directly in the BDI programming paradigm. Coupling BDI systems with Agent Based Modelling and Simulation (ABMS) systems, one can create powerful simulations that can be used for a range of analysis, training and community education purposes.