2 Criteria for a Good Model of Agent Learning

2.1 Rigour

Whatever the uses to which they are put, we take the view that simulation models should not be less rigorous than analytic models. Rigour provides a set of rules according to which developments and meaning of the models can be assessed. In effect, models provide the basis for a discourse about as-yet-unrealized outcomes and, to be useful, we believe that such discourse must also be disciplined.

Our primary concern then is to devise a formal representation of learning by agents which is both general and rigorous. In particular cases, representations of learning which conform to our formal representation could doubtless generate equilibrium results in appropriate models. More general models which entail learning by agents will inform discussions of, for example, strategic decision-making processes which are not expected by agents to result in steady-state outcomes.

Modelling Learning as Modelling - 23 FEB 98

[Next] [Previous] [Up] [Top] [Contents]

Generated with CERN WebMaker