Invited Speakers

Prof. Flaminio Squazzoni

Department of Social and Political Science 

University of Milan

Exploring multiple generative paths for aggregate dynamics with agent-based models

In this lecture, I shall propose the use of agent-based models as a tool for exploring multiple generative paths for complex aggregate dynamics which could be overlooked with statistical models estimating parameters from data. I will offer two examples: (I) a model of trust dynamics calibrated on data on individual choices in repeated strategic games played in the laboratory, where the model was used to vary initial network configurations, explore boundary conditions and perform counter-factual test; (II) a model of advise networks where a previous Stochastic Actor-Oriented Model fitted on empirical data was agentized to discover potential equifinality outcomes which were hidden in parameter estimation. While the computational social science research programme puts a great emphasis on the power of data and model parameter estimation, here I will propose a ‘many models’ micro-macro generative approach that explicitly avoids any ‘causal exclusivism’, helps to integrate models and data, and uses simulation as an experimental environment.