Social-Behavioral Modeling for Complex Systems
by Paul K. Davis, Angela O'Mahony, Jonathan Pfautz
25 Causal Modeling with Feedback Fuzzy Cognitive Maps
Osonde Osoba1 and Bart Kosko2
1 RAND Corporation and Pardee RAND Graduate School, Santa Monica, CA 90401, USA
2 Department of Electrical Engineering and School of Law, University of Southern California, Los Angeles, CA 90007, USA
Introduction
Social scientific theories can benefit from computational models even when the relevant social scientific concepts and phenomena can be hard to quantify. This chapter describes a powerful tool for modeling causal relationships: fuzzy cognitive maps (FCMs). FCMs offer a practical and flexible way to model and process the interwoven causal structure of policy and decision problems.
Existing tools for causal modeling include Bayesian belief networks (BBNs) and system dynamics (SD) models. FCMs have some notable advantages over the default models of causality: FCMs model causal loops naturally, they combine separate knowledge sources into a single FCM, and they permit fast pattern inference and data‐driven adaptation at low computational cost. FCMs also lend themselves particularly well for modeling ambiguous elicited causal beliefs because of their use of fuzz. A flexible and robust modeling approach for capturing causal beliefs is important for social and behavioral models since beliefs are foundational to many observed behaviors.
The next sections describe FCMs briefly. We conclude with two FCM policy applications. The first shows how FCMs can assist in modeling the many causal ...