The analogical roots of agent-based modeling in economics and social sciences: the case of innovation dynamics

Agent-based modeling (ABM) is a simulation technique which has been increasingly integrated into the economic discipline in order to understand complex systems. However, most of everyday research activities rely on the researchers' consensus concerning practical choices about modeling strategies, computational boundaries under scrutiny and the extent of empirical validation. Particularly lacking are reflections on the semantic construction of conceptual models. The paper reviews existing theoretical frameworks leading to the understanding of ABM as a technique where the cognitive processing instantiated by the instrument is distributed across different modeling layers, including conceptual, algorithmic and computational. These layers can be interpreted as an interlinked set of analogies. Then, the paper introduces a framework for assessing ABM conceptual adequacy and tests it on two families of models in the economics of innovation field.