Behavioral analysis of fuzzy cognitive map models by simulation

Fuzzy Cognitive Maps (FCM) are widely applied to support decision making and for the prediction of the future behavior of systems. In this paper several real expert-defined models of an organization will be investigated. Simulations were performed in order to examine the asymptotic behavior of these models, especially to find (all) fixed point attractors, or to discover chaotic behavior. A sigmoid type of inference function was applied, and the effect of the steepness parameter on simulation results was investigated, too. It will be shown by several examples that the steepness parameter may even change the number and values of the final state vectors. In the end, the systematic study of the behavior of the models in terms of the changes of the (uncertain) relationship values among system components were studied.